Author: Security – Computerworld

Blood in the iPhones? Apple faces criminal charges

Apple has been accused of knowingly financing a trade characterized by atrocities in criminal litigation launched by the Democratic Republic of Congo (DRC). It relates to use of so-called “blood minerals” in its devices.

While Apple will be the first tech firm to be targeted, it is unlikely to be the last, given that every digital device makes use of these materials.

The accusations reflect an unfortunate truth: that to some extent, the digital world and the devices used to drive it are built on slavery and other forms of human misery.

That’s the take-home message at the heart of the litigation. It claims Apple’s supply chain is contaminated by “blood minerals” (also called “conflict minerals”) sourced in sub-Saharan Africa. Criminal complaints have been filed against subsidiaries of Apple in France and Belgium. In its complaint, the DRC argues that by enabling such contamination within its supply chain, Apple is contributing to atrocities taking place across the region.

What exactly are blood minerals? 

They are rare materials, in this case including tin, tantalum, and tungsten, that are extracted by various forms of coerced labor, including slave and child labor, at mines controlled by armed groups in eastern Congo. These materials are in enormous demand across tech.

To understand the vast suffering generated by the hugely profitable trade, it is important to note that the region has an estimated $24 trillion value of these materials remaining to be tapped — and also has one of the world’s poorest populations. Blood minerals fuel instability and conflict, with armed groups using the profits to purchase weapons, creating a never-ending cycle of violence and misery. Millions have died as a result of the ongoing conflict. This is far from being a new problem

While attempts have been made to mitigate this trade by use of various verification schemes, it remains highly possible that at least one and possibly all of the electronic objects that you use contain at least some minerals sourced from this exploitation.

After all, tin, tantalum, and tungsten are used in solder, semiconductors, batteries, memory, RF filters, and microprocessors, all of which you’ll find in almost any digital device.

Most manufacturers have adopted some form of policy concerning the trade. In fact, the denials of any involvement in the mining and use of conflict minerals are so strong you’d be forgiven for wondering how the trade actually makes any money. But that plausible deniability may not be all it seems.

Apple perhaps not the biggest culprit

While it is the whipping horse for this litigation, Apple may not be the biggest culprit. Eager to be recognized for its “values,” Apple has made big commitments to responsible sourcing, saying that while it doesn’t source materials directly, it does require its suppliers to source materials responsibly.

The company’s Standards for Responsible Sourcing of Materials are based on leading international guidance, including the United Nations Guiding Principles on Business and Human Rights and the Organisation for Economic Co-operation and Development (OECD) Due Diligence Guidance for Responsible Supply Chains of Minerals from Conflict-Affected and High-Risk Areas

“In 2023, 100 percent of the identified tin, tantalum, tungsten, gold (3TG), cobalt, and lithium smelters and refiners in Apple’s supply chain completed assessments to verify compliance with our standards,” the company said last year. “We work with third-party audit programs at an industry-wide level to identify environmental and social, and governance risks at the smelter, refiner, and mining levels.”

The company also sits on the steering committee of the Responsible Minerals Initiative (RMI).

Apple will presumably soon publish its 2024 Conflict Minerals Report, as required by the SEC. Its last report covering 2023 appeared in March 2024. In it, Apple confirmed that it removed 14 smelters and refiners unwilling to face an audit to ensure lack of involvement in the conflict minerals trade. The company has cut ties with 25 manufacturing supplier facilities and 231 suppliers since 2009.

“We view removing a supplier from our supply chain as a last resort, because in our experience, it does not provide workers with needed remedy and could allow violations to continue elsewhere in the industry,” Apple said in its report. 

“Based on our due diligence efforts, including analyzing the information provided by third-party audit programs, upstream traceability programs, and our suppliers, we found no reasonable basis for concluding that any of the smelters or refiners of [tin, tungsten, tantalum, and gold] determined to be in our supply chain as of December 31, 2023, directly or indirectly financed or benefited armed groups in the DRC or an adjoining country,” that report said.

A reliance on systemic wrongdoing?

The DRC’s litigation doesn’t buy some of Apple’s arguments, claiming instead that the company is using them to mask the damage it is doing.

In a press release supplied by Amsterdam & Partners, the litigants claim Apple uses minerals laundered through international supply chains. It is also accused of “using deceptive commercial practices to assure consumers that the tech giant’s supply chains are clean.”

Reuters reports that the complaint filed in France states, “It is clear that the Apple group, Apple France and Apple Retail France know very well that their minerals supply chain relies on systemic wrongdoing.”  

The claim points to Apple’s use of the International Tin Supply Chain Initiative (ITSCI), arguing that Apple is using the discredited monitoring and certification scheme to falsely claim its supply chain is clean. The RMI — a group Apple helps steer — rejected ITSCI two years ago.

The criminal complaint cites research from the United Nations, US State Department, and international NGOs that shows an extensive laundering enterprise through the illegal trade in conflict minerals sourced from Congolese territory. 

“These organizations have demonstrated the dependent nature of relationships between perpetrators of this looting and some of the biggest producers of consumer electronics, such as mobile telephones and computers, and companies in the automotive, aviation and renewable energy sectors,” the press release explains.

A need for a united front

That may be true, and Apple may be able to defend itself using the same argument. 

To truly combat this trade, a true industry-wide commitment must be reached, regulated or self-regulated. In its absence, conflict laundering will continue to be a problem. 

Paula Pyers, Apple’s then Senior Director of Supply Chain Social Responsibility, said as much in 2017: “If more companies do not come to the table to press for change through their own supply chains, particularly in the absence of regulation, the types of systemic change we are all seeking are frankly not going to occur.”

Ultimately, the situation is hard to clarify, in part because alongside the conflict minerals there are also thousands of informal small-scale mines and miners, with some of the world’s poorest people taking great risks to make some money. Corruption and instability at some otherwise ‘clean’ mining sites means conflict minerals can still get into the system, making it next to impossible to deliver a cast-iron guarantee.

The inability to make that guarantee forms part of the argument the DRC is making in this case.

However, it also forms part of Apple’s counter-argument, giving it the power to say that it is already doing everything it can to combat the trade using what resources it has available.

The truth of that argument will be for the courts to decide.

Putting them out of business

The biggest way to prevent any illicit trade is simply to stop using the products based on it.

And when it comes to electronics, one way to do that is to use recycled minerals. That’s precisely what Apple is doing, and fast — rapidly replacing these minerals with recycled materials.

Its most recent Environmental Progress Report promises that cobalt, tin soldering, gold plating, and rare earth elements will all be 100% recycled by 2025. Already, “more than 99 percent of the tungsten in our products comes from recycled sources,” the company said.

While the criminal lawsuit against Apple may generate problems, it may also give the company a podium from which to promote the need for a more unified approach to policing the trade in blood minerals. It will certainly give it a pulpit from which to preach its move to recycled components and work toward a circular manufacturing system. 

What the truth of the matter turns out to be will be for the courts to decide.

You can follow me on social media! Join me on BlueSky,  LinkedInMastodon, and MeWe

Analyze text using natural language with Claude for Google Sheets

You can do a lot more with generative AI (genAI) in Google Sheets than get help writing spreadsheet formulas. Thanks to extensions such as Claude for Sheets, you can also do things like:

  • Determine whether text sentiment is positive, negative, or neutral
  • Classify text into categories
  • Extract email addresses, phone numbers, and other entities from plain text

And you can do this without having to write computer code in a language like Python.

Instead, you only need spreadsheet formulas as simple as:

=claudeExtract("sentiment of positive, negative, or neutral", A2)

Let’s see how to implement some of these use cases with the help of Claude for Sheets, an add-on for Google Sheets offered by Anthropic, maker of the Claude family of large language models (LLMs).

Just remember:

  1. LLMs aren’t always accurate. The more important your task, the more effort you should put into checking Claude’s results.
  2. Think twice about sending sensitive data to an LLM. If it’s corporate data, follow company policies.
  3. Claude for Sheets costs money to use. It’s a pretty trivial amount for moderate amounts of text, but understand the pricing if you’ve got a massive data set. (Although for a lot of data, performance will be slow and you’re probably better off with another solution.) More on pricing in a bit.

Claude for Sheets setup

You’ll need two things in addition to a Google account in order to run Claude for Sheets: an Anthropic API key and the Claude for Sheets extension.

You can get an Anthropic API key by creating a free account. Once you log in, your dashboard home page should show an option to create API keys.

screenshot of anthropic dashboard home page

You can request an API key from the Anthropic dashboard home page.

Sharon Machlis / IDG

Click the button to create a key and heed the warning to copy your API key (just a string of characters) when it appears — once you close that dialog box, you won’t be able to access it again on the Anthropic website. Store the key securely, just as you would a password.

You’ll likely need to add some money to your Anthropic account before you can use the Claude API. You shouldn’t need much unless you’re uploading a lot of data. (I added $20 months ago and still have more than half left — and I’ve coded public-facing apps using Claude in addition to playing with Sheets.)

To add Claude for Sheets to your Google account, download it from the Google Workspace Marketplace and install it.

You need to set up Claude for Sheets in each spreadsheet where you want to use it. Create a new Google spreadsheet and go to Extensions > Claude for Sheets > Open Sidebar. Then click on the sidebar’s hamburger menu (three horizonal lines), choose Settings, and click on API provider. You want to choose Anthropic. Enter your API key where it says the key is missing.

srcset="https://b2b-contenthub.com/wp-content/uploads/2024/12/claude_for_google_sheets_02_enter_key.jpg?quality=50&strip=all 976w, https://b2b-contenthub.com/wp-content/uploads/2024/12/claude_for_google_sheets_02_enter_key.jpg?resize=257%2C300&quality=50&strip=all 257w, https://b2b-contenthub.com/wp-content/uploads/2024/12/claude_for_google_sheets_02_enter_key.jpg?resize=768%2C896&quality=50&strip=all 768w, https://b2b-contenthub.com/wp-content/uploads/2024/12/claude_for_google_sheets_02_enter_key.jpg?resize=877%2C1024&quality=50&strip=all 877w, https://b2b-contenthub.com/wp-content/uploads/2024/12/claude_for_google_sheets_02_enter_key.jpg?resize=597%2C697&quality=50&strip=all 597w, https://b2b-contenthub.com/wp-content/uploads/2024/12/claude_for_google_sheets_02_enter_key.jpg?resize=144%2C168&quality=50&strip=all 144w, https://b2b-contenthub.com/wp-content/uploads/2024/12/claude_for_google_sheets_02_enter_key.jpg?resize=72%2C84&quality=50&strip=all 72w, https://b2b-contenthub.com/wp-content/uploads/2024/12/claude_for_google_sheets_02_enter_key.jpg?resize=411%2C480&quality=50&strip=all 411w, https://b2b-contenthub.com/wp-content/uploads/2024/12/claude_for_google_sheets_02_enter_key.jpg?resize=308%2C360&quality=50&strip=all 308w, https://b2b-contenthub.com/wp-content/uploads/2024/12/claude_for_google_sheets_02_enter_key.jpg?resize=214%2C250&quality=50&strip=all 214w" width="877" height="1023" sizes="(max-width: 877px) 100vw, 877px">

Enter your Anthropic API key after opening the Claude for Sheets sidebar.

Sharon Machlis / IDG

Next, scroll down and choose a default model. Anthropic offers several different models under the Claude umbrella, and their capabilities and pricing vary considerably.

Haiku is the smallest, fastest, least expensive, and least capable model offered in Claude for Sheets. I’d start there while experimenting.

srcset="https://b2b-contenthub.com/wp-content/uploads/2024/12/claude_for_google_sheets_03_choose_model.jpg?quality=50&strip=all 729w, https://b2b-contenthub.com/wp-content/uploads/2024/12/claude_for_google_sheets_03_choose_model.jpg?resize=253%2C300&quality=50&strip=all 253w, https://b2b-contenthub.com/wp-content/uploads/2024/12/claude_for_google_sheets_03_choose_model.jpg?resize=587%2C697&quality=50&strip=all 587w, https://b2b-contenthub.com/wp-content/uploads/2024/12/claude_for_google_sheets_03_choose_model.jpg?resize=142%2C168&quality=50&strip=all 142w, https://b2b-contenthub.com/wp-content/uploads/2024/12/claude_for_google_sheets_03_choose_model.jpg?resize=71%2C84&quality=50&strip=all 71w, https://b2b-contenthub.com/wp-content/uploads/2024/12/claude_for_google_sheets_03_choose_model.jpg?resize=405%2C480&quality=50&strip=all 405w, https://b2b-contenthub.com/wp-content/uploads/2024/12/claude_for_google_sheets_03_choose_model.jpg?resize=303%2C360&quality=50&strip=all 303w, https://b2b-contenthub.com/wp-content/uploads/2024/12/claude_for_google_sheets_03_choose_model.jpg?resize=211%2C250&quality=50&strip=all 211w" width="729" height="865" sizes="(max-width: 729px) 100vw, 729px">

Choose a default model to use in Claude for Sheets. Haiku is the smallest and least expensive.

Sharon Machlis / IDG

As of this writing, Haiku costs 80 cents per million tokens in and $4 per million tokens out. One million tokens is roughly 4 million characters or 750,000 English words. The priciest and most powerful model, Opus, is $15 per million input and $75 per million output. The middle model, Claude 3.5 Sonnet, is $3/M in and $15/M out.

Whatever you choose as the default model for a spreadsheet, you can override it by specifying a different model in the formulas you create, as we’ll see in the examples below.

Case 1: Sentiment analysis

Start by entering the text you want to analyze in your A column. I entered the following five sample “review” text blurbs into my spreadsheet’s A column (ReviewText), one item per row:

ReviewTextSentiment
The new iPad Mini 7 has a great display, speedy response, and is light enough to hold for long periods of time. It’s a significant upgrade from my old (ancient?) iPad 5. My lone nit is that the form factor is thicker and less sleek now due to the change in connectors, but I’m still glad I upgraded. 
This desktop computer has a better processor and can handle much more demanding tasks such as running LLMs locally. However, it’s also noisy and comes with a lot of bloatware. 
The charger is affordable and does exactly what I want. 
This charger doesn’t work like I expected. 
This charger seems like a great value – until you try to use it. The connection is flakey and my device often ends up not fully charged overnight. 

In column B (Sentiment) — more specifically, in cell B2 next to my first text item, I added the formula

=claudeExtract("sentiment analysis of 'positive', 'negative', or 'neutral’", A2, "claude-3-5-haiku-latest", "temperature", 0)

And then I clicked and dragged the formula down the rest of the rows next to cells with text.

Here’s an explanation of the formula.

=claudeExtract uses the syntax claudeExtract(property, text, model, argumentValuePairs)

Property is defined in the help file as “The property to extract from the text, e.g. ‘phone’.” For this case, I used a property of “sentiment analysis of ‘positive’, ‘negative’, or ‘neutral’” since that’s what I want Claude to extract from my text.

Text is my source cell location. If my first row of text is in cell A2, text value will be cell A2. Clicking and dragging the formula down the rest of the column applies it appropriately to the other text cells.

Although I chose a default model in Claude for Sheets main settings, I specified a model in the formula too so I could add optional arguments at the end of the formula. Google Sheets formulas don’t like it if you skip expected parameters when adding more at the end. Anthropic’s Model names documentation helpfully lists the specific syntax to use in formulas for each of its available models. In this case I chose claude-3-5-haiku-latest — which specifies the latest available version of the Claude 3.5 Haiku model.

The one optional argument I added was to set the model’s temperature to 0. An LLM’s temperature tells it how much randomness to use. Should it opt for the most likely result again and again (low temperature) or try to mix things up a bit (higher temperature)?

It’s sometimes helpful to view temperature as a measure of model “creativity.” Higher temperature can be nice when chatting with an LLM, so it doesn’t sound robotic and repetitive. However, for technical tasks, “more likely to be accurate” is a better idea than “be creative,” and that means lower temperatures. Anthropic advises: “For multiple-choice or analytical tasks, you’ll want [temperature] close to 0.”

Note: This ability to set model parameters like temperature is one advantage of using the LLM’s API via a spreadsheet or computer programming instead of a free chatbot without granular temperature control.

Results

Below is Claude’s sentiment analysis on my sample text. I agreed with all of them:

Haiku: Sentiment analysis

ReviewTextSentiment
The new iPad Mini 7 has a great display, speedy response, and is light enough to hold for long periods of time. It’s a significant upgrade from my old (ancient?) iPad 5. My lone nit is that the form factor is thicker and less sleek now due to the change in connectors, but I’m still glad I upgraded.positive
This desktop computer has a better processor and can handle much more demanding tasks such as running LLMs locally. However, it’s also noisy and comes with a lot of bloatware.neutral
The charger is affordable and does exactly what I want.positive
This charger doesn’t work like I expected.negative
This charger seems like a great value – until you try to use it. The connection is flakey and my device often ends up not fully charged overnight.negative

Two important tips:

If you see errors or delays in some cells, you can ask Claude for Sheets to recalculate those by opening the Claude extension’s dropdown menu (Extensions > Claude for Sheets) and clicking on Recalculate all “DEFERRED / THROTTLED / #ERROR! Cells.

google sheets menu for recalculating claude for sheets cells

Claude may not finish all your spreadsheet calculations in one pass. If that happens, you can ask it to recalculate cells with missing values via the Extensions menu.

Sharon Machlis / IDG

To keep results from updating — which might change them and costs more API money — copy and paste the column values only (excluding the formulas) to a new column and then delete the original column with formulas. Anthropic says Claude for Sheets results are cached for a week, so opening the spreadsheet again after a week may cause everything to recalculate.

Case 2: Text classification into categories

For this test, I took a few social media posts about the R programming language (often used for data analysis), genAI tools, and other topics — some modified to try to make the topics slightly less obvious — and asked Claude to categorize them as Data, Generative AI, Security & Privacy, or Other.

I used the same workflow as above, just changing the formula’s property text using natural language instructions:

=claudeExtract("one or more categories among 'Data', 'Generative AI', 'Security & Privacy'. If more than one apply, include all. If none apply, say 'Other'.", A2, "claude-3-5-haiku-latest", "temperature", 0)

Results weren’t bad but were imperfect. The Haiku model didn’t apply both Security & Privacy and Generative AI to the text about LLM vulnerabilities, and it didn’t know that R involves data.

Haiku: Text categorization

Kyle has some great resources for using the GIS data sets he mentions, especially “Analyzing US Census Data: Methods, Maps, and Models in R” and the tidycensus packagedata
OpenAI today announced new model 4o — basically 4o preview out of preview — it says is better and faster than the preview version. And multimodal. Also a new “pro” tier for ChatGPT and its 4o model, at an eye-popping $200/month.generative ai
An Introduction to R is a free online book by several professors (Univ of Aberdeen, Danish Technical Univ) and R practitioners that aims to, well, introduce you to using R. If you teach R, feel free to use some or all of the content in this book [with attribution]”. https://intro2r.comother
Prompt injection and supply chain vulnerabilities remain the main LLM vulnerabilities but as the technology evolves new risks come to light including system prompt leakage and misinformation. https://www.csoonline.com/article/575497/owasp-lists-10-most-critical-large-language-model-vulnerabilities.htmlsecurity & privacy
This story “tracked thousands of US military & intel personnel coming & going from classified sites, incl. NSA hubs & nuclear vaults. We know where they sleep, what they eat, and which brothels they visit. It’s an ocean of blackmail & national secrets within reach of every spy agency in the world.”data,security & privacy
“Logging on to Bluesky for the first time felt a bit like walking into your apartment after a professional deep clean. The layout is the same, as are the appliances and furniture. But it just felt good to be there. I wanted to hang out.”other

When I tried with the Sonnet model (by swapping in claude-3-5-sonnet-latest as the model name), it did apply both categories to the LLM security text but also couldn’t recognize the R blurb as being data-related. The Opus model (claude-3-opus-latest) didn’t know R is data-related either.

My conclusion on categorizing text after this and other experiments: Either the text needs to be fairly straightforward about topics, or the prompt should include specific instructions about phrases likely to show up in your data that the LLM doesn’t know how to classify. In other words, experiment a bit and add examples or more detailed explanations when needed. When I added “Text about the R programming language is topic Data.” to the end of my prompt’s property text, Haiku applied “data” to my R-related posts.

Case 3: Entity extraction

I used some sample text to test extracting email addresses and phone numbers. The text was originally generated by a chatbot (Perplexity), and I modified it a bit. I then added that text to a new tab on the spreadsheet, and inserted the following formulas in columns B and C:

=claudeExtract("all phone numbers", A4, "claude-3-5-haiku-latest", "temperature", 0)
=claudeExtract(“all email addresses", A4, "claude-3-5-haiku-latest", "temperature", 0)

Haiku performed quite well on this one:

Haiku: Phone number and email address extraction

TextPhoneEmail
The team is at marketing@company.com and (914) 737-9938. Headquarters is @ 512 Main St.914-737-9938marketing@company.com
John Smith from accounting (j.smith@business.net) will be out of office until Friday. For urgent matters, reach him at 888-555-9012.888-555-9012j.smith@business.net
Contact support: help_desk@tech.org or 1-800-555-3456 Technical issues? sarah.tech@company.com For appointments: +1 (444) 867-53091-800-555-3456, +1 (444) 867-5309help_desk@tech.org, sarah.tech@company.com
Meeting notes from yesterday: – Dave (d.wilson@dept.com) will handle the project – Marketing team contact: marketing@company.com / 555.777.8888 – Emergency line: 911-555-0123555.777.8888, 911-555-0123d.wilson@dept.com, marketing@company.com
Please forward any invoices to billing@finance.com or fax to (777) 555-0147777-555-0147billing@finance.com

Trying to extract company names was tougher, since it’s a lot easier to spot phone number and email address patterns than to know the difference between a company name and a product name.

I tested company and product extraction with some recent headlines and summaries from Computerworld’s home page, asking simply for “companies” and “products” in my formulas. Haiku’s results:

Haiku: Company and product name extraction

TextCompaniesProducts
10 steps to smarter Google account security Give yourself some added peace of mind by giving your Google account a thorough set of security reinforcements, both on Android and your desktop.GoogleGoogle account, Android, desktop
How ChatGPT works with iOS 18.2 iPhones and Siri Apple Intelligence is about to receive its first important update — introducing ChatGPT access.ChatGPT, iOS, Apple, SiriChatGPT, iOS 18.2 iPhones, Siri, Apple Intelligence
OECD: GenAI is affecting jobs previously thought safe from automation Though the technology will likely lead to new jobs, they may not benefit those who lost work due to automation.OECDGenAI
Microsoft moves to stop M365 Copilot from ‘oversharing’ data The generative AI assistant can surface sensitive information in over-permissioned files, a growing concern for businesses testing the technology. Microsoft is adding new features to SharePoint and Purview to make it easier to control what the tool can access.MicrosoftMicrosoft 365 Copilot, SharePoint, Purview

It was a bit disappointing that Haiku tagged ChatGPT, iOS, and Siri as companies. Sonnet did better on that one overall but worse on the last listing, adding two products as companies along with Microsoft:

Sonnet: Company and product name extraction

TextCompaniesProducts
10 steps to smarter Google account security Give yourself some added peace of mind by giving your Google account a thorough set of security reinforcements, both on Android and your desktop.GoogleGoogle account
How ChatGPT works with iOS 18.2 iPhones and Siri Apple Intelligence is about to receive its first important update — introducing ChatGPT access.AppleiOS 18.2, iPhones, Siri, ChatGPT
OECD: GenAI is affecting jobs previously thought safe from automation Though the technology will likely lead to new jobs, they may not benefit those who lost work due to automation.OECDGenAI
Microsoft moves to stop M365 Copilot from ‘oversharing’ data The generative AI assistant can surface sensitive information in over-permissioned files, a growing concern for businesses testing the technology. Microsoft is adding new features to SharePoint and Purview to make it easier to control what the tool can access.Microsoft, SharePoint, PurviewM365 Copilot, SharePoint, Purview

When I tried the pricey Opus model, it got rows 2-4 correct but added Android as a company on the first row. My conclusion: I either need to give these models better prompts and more examples for a task like this or tolerate some inaccuracies.

However, models continue to improve, and a task that’s beyond their capabilities now may work better a few months down the road.

For the sake of comparison, I gave all this text to OpenAI’s new o1 model in the ChatGPT Plus chatbot and asked it to extract products. Those results were better: o1 correctly identified Google account, Android, ChatGPT, iOS 18.2, iPhones, Siri, M365 Copilot, SharePoint, and Purview as products.

And it was the only model to extract just Google, Apple, and Microsoft as “companies” — OECD, the Organisation for Economic Co-operation and Development, is not technically a company. When I asked o1 to extract both companies and organizations, it responded: Google (company), Apple (company), OECD (organization), and Microsoft (company).

(OpenAI doesn’t have its own extension for Google Sheets, although there are some paid third-party applications.)

Meanwhile, though, if you can tolerate some imperfect results and less than top-speed performance, you can start doing LLM-based natural language processing right within a spreadsheet. As Ethan Mollick, author of Co-Intelligence: Living and Working with AI, advised on Bluesky recently:

“I think firms worrying about AI hallucination should consider some questions:

1) How vital is 100% accuracy on a task?

2) How accurate is AI?

3) How accurate is the human who would do it?

4) How do you know 2 & 3?

5) How do you deal with the fact that humans are not 100%?

Not all tasks are the same.”

Making AI popular is a marathon, not a sprint

AI is everywhere, but from where I sit, there are some strong signals that suggest the road to AI Everywhere is going to be a long-distance endurance race, rather than a sprint.

While these tools are seeing a lot of use since ChatGPT burst upon the scene, they are also generating lots of cost — and despite all the marketing, it still isn’t really clear if consumers are buying hardware based on ‘AI Inside.’ (Though it is already clear that workers are using AI in the shadows).

Even iPhone users — usually the fastest adopters of cutting-edge tech — seem to want convincing that AI is all it’s cracked up to be.

It’s almost as if people purchasing these products are a little turned off by a technology that threatens to destroy their employment, exacerbate wealth inequalities, and supercharge surveillance advertising in exchange for email summaries and a search engine powerful enough to help you file your next welfare benefit claim. 

Who could have seen that coming?

When it comes to tech, consumers have developed a resistance to the new. They’ve seen both bank accounts and politics hacked by technologies originally sold to them under the promise of making things better, and they are annoyed that the ad-free streaming television they began paying for now comes with added ads. They’ve seen this happen again and again, as a result of which they are suspicious of new tech — even when it is private and trusted like the AI that Apple is promising to provide

Why do I think this?

A recent survey of 2,000 US smartphone users by trade-in site SellCell found that 73% of iPhone users and 87% of Samsung users say the AI features they have been provided with in the latest software updates are adding little or no value. Samsung introduced its first genAI smartphone in March this year, while Apple rolled out Apple Intelligence with iOS 18.1 in October.

The survey was taken before Apple introduced iOS 18.2, and while it can’t be seen as representative, I think it shows that the expectation that AI will somehow deliver a big bump in device and PC sales may be misplaced — though server sales will see a big spike as service providers and businesses implement AI in their systems.

This doesn’t mean iPhone users aren’t interested in AI. Among iOS users who responded to the SellCell survey, 47.6% called it “key” in choosing a new phone; just 23.7% of Samsung users felt the same way. It may also matter that 21.1% of smartphone users already see AI as a very important deciding factor when choosing a new device.

What are the most popular AI features on phones?

While the survey does show there’s some journey to go before the promise of AI resonates fully with its audience, it also reveals which of the features made available in iOS 18.1 (pre-Genmoji) most interested users: Writing Tools (72%), Notification summaries (54%), Priority Messages (44.5%), Clean Up in Photos (29.1%), and Smart Reply in Mail and Messages (20.9%).

For comparison, Samsung AI users checked out thusly: Circle to Search (82.1%), Photo Assist (55.5%), Chat Assist (28.8%), Note Assist (17.4%), and Browsing Assist (11.6%).

It’s clear that Apple’s decision to Sherlock Grammarly (and do it privately) has given the company its most attractive suite of AI tools. People like tools that help them do everyday things better, it seems.

That desire for enhanced productivity is also what is driving employees to use AI services for their work, sometimes to the detriment of security policies and customer privacy.

At this stage in AI deployment, consumers still need convincing, and companies still need time to think about how best to deploy the tech — though a recent CCS Insight survey of business leaders showed that 82% are in the process of deployment. We are early in the mass adoption curve, and in that environment, taking a cautious and deliberate stance to adoption seems to be the best way to avoid falling into any unexpected disasters. Which is, oddly enough, how Apple has been approaching the topic since the get-go.

Will you pay for AI?

While I don’t see platform-wide AI as anything like the same animal as the fast-growing assemblage of customized focused AI services for specific industries and tasks, it still feels like the opportunity to monetize general purpose mass-market AI services remains some way off. 

Apple users are a little more ready. 

They are a lot more likely to consider investing in AI subscriptions (which may yet justify OpenAI’s gamble to offer its services through Apple’s kit), but subscription is a trickle, rather than a flood. The survey tells us 11.6% of them are likely to pay for a subscription to use AI services, in contrast to just 4% of Samsung users. 

However, most smartphone users (86.5% of iPhone users and 94.5% of Samsung users) said they would not pay to use AI. 

The decision seems pretty polarized in that just 1.9% of Apple users aren’t sure if they’d pay or not — in other words, there’s a pretty clear division between the two sides. That means the task of convincing people to adopt is already entirely based on showing those refuseniks compelling usage cases that help them come to terms with, trust, and develop the desire to use AI in some way.

Gathering the tribes

Eroding this resistance and bestowing these services with an increased perception of value will clearly be part of the product management journey for Apple AI, and every other kind of artificial intelligence. It may well be that as the uses of these tools become more widespread, people’s enthusiasm will increase. It is worth noting that at some point prior to the introduction of Apple Intelligence, some reports were claiming that some inside Apple themselves remained uncertain if the first echelon of tools to ship would resonate with consumers.

They may find themselves slightly relieved that while AI features don’t seem to grab the attention of every iPhone user, the ones who are making use of the tools seem pretty enthusiastic. iPhone users are also trying the new AI services as they appear, with 41.6% of those owning an iPhone that supports them confirming they’d used them, mostly writing tools, notification summaries, and priority messages.  

What does this tell us? 

I think it tells us that the current trend of just waving at a gadget that happens to support genAI and calling it an “AI smartphone” or an “AI PC” — or, in Apple’s case, a “complete mobile to PC AI ecosystem” — may need to be refined.

People need to feel more trust in these solutions, want to be convinced that potential damaging use is mitigated against, and most of all want applied use cases in which the tech can be applied in positive, life-enhancing ways. It’s also possible that the hype of AI is already behind us, and that now the search must shift to identifying those compelling instances in which the tech addresses significant human need

You can follow me on social media! Join me on BlueSky,  LinkedInMastodon, and MeWe

What IT hiring looks like heading into 2025

Despite low unemployment in the IT industry, IT hiring has slowed over the past two years and is expected to plateau in 2025 as employers prioritize experienced candidates with specialized skills, according to three new studies. They found that companies are shifting from volume hiring to “quality hiring,” leading to fewer roles for tech job candidates.

Staffing firm ManpowerGroup, which just published its Q4 report on hiring, claimed IT hiring leads all other professions in the US but predicted employers will still pull back on hiring next year because of “economic uncertainty.”

“As we move into 2025, we’re seeing stable year-over-year hiring trends, with employers holding onto the talent they have and planning muted hiring for the quarter ahead,” said Jonas Prising, ManpowerGroup Chair & CEO.

Overall, the studies by ManpowerGroup, online hiring platform Indeed, and Deloitte Consulting showed that IT hiring will increasingly be based on having flexible skills that can meet changing demands. “Employers know a skilled and adaptable workforce is key to navigating transformation, and many are prioritizing hiring and retaining people with in-demand flexible skills that can flex to where demand sits,” Prising said.

Becky Frankiewicz, president of the North America Region at ManpowerGroup, said that while hiring plans remain steady, a closer look at the data indicates employers will adopt a more strategic approach in the first quarter of 2025, with a focus on retaining existing talent and prioritizing in-demand skills in areas like IT, financial services, and manufacturing, she said.

“Real time, we are seeing companies take longer to make a hire — and seasonal hiring patterns are changing too, [with] more hiring earlier on in the season and less intense [during] holiday periods,” Frankeiwicz said. “This tells us this isn’t your typical cycle. We’re watching the labor market normalize at different speeds across industries. It’s a sign that organizations are adapting and becoming more precise in their workforce planning.” 

IT Jobs and AI

ManpowerGroup

Indeed economist Cory Stahle said the latest US jobs data shows increased IT hiring, indicating stability heading into the new year. At the same time, Stahle reiterated past assessments by others of a potential cooling, as companies continue to rebalance after a hiring surge in 2021 and early 2022 driven by pandemic-related demand for tech services.

“…Knowledge worker roles like in the tech and marketing sectors are cooling,” Indeed’s 2025 US Jobs & Hiring Trends Report said.

Quit rates also reflect labor market dynamics and job seeker confidence. A declining quit rate could signal lower enthusiasm, as workers are less confident in finding new jobs, according to Indeed. After rising sharply during the “Great Resignation” of 2021-2022, the quits rate has steadily declined, Indeed said. In September, it hit 1.9%, its lowest since July 2015, before rising to 2.1% in October.

The pullback in job postings, particularly in tech, likely affects job seekers’ confidence, Indeed said.

Additionally, generative artificial intelligence (genAI) is affecting entry-level hiring, as more and more repetitive or lower-level jobs, such as helpdesk tasks, get automated. Those jobs can often be handled by AI agents such as Microsoft 365 Copilot, Amazon CodeWhisperer, and IBM Watson Assistant.

Job postings for GenAI skills

Indeed

“Right now, genAI appears to have the greatest potential impact on knowledge workers,” Stahle said. “Recently released Hiring Lab research shows that genAI often does well with theoretical tasks like creating a recipe and does less well at hands-on tasks like making food.”

AI isn’t replacing jobs so much as it is reshaping the nature of work, said Elizabeth Lascaze, a principal in Deloitte Consulting’s Human Capital practice. She, too, sees evidence that entry-level roles focused on tasks like note-taking or basic data analysis are declining as organizations seek more experienced workers for junior positions.

“Today’s emerging roles require workers to quickly leverage data, generate insights, and solve problems,” she said, adding that those skilled in using AI, such as cybersecurity analysts applying AI for threat detection, will be highly sought after.

Although the adoption of AI has led to some “growing pains,” many workers are actually excited about it, Lascaze said, with most employees believing it will create new jobs and enhance their careers. “Our survey found that just 24% of early career workers and 14% of tenured workers fear their jobs will be replaced by AI,” Lascaze said. “Tenured workers are more likely to lead organizational strategy, so they may prioritize AI’s potential to improve efficiency, sophistication, and work quality in existing roles rather than AI’s potential to eliminate certain positions.

“These workers reported being slightly more focused on building AI fluency than early-career employees,” Lascaze said. “With so many early-career employees reporting excitement around using AI, organizations should establish reverse mentorship opportunities where tenured staff can learn the ropes.”

GenAI-related jobs are still rare, accounting for about two in 1,000 nationwide as of October. But they are growing quickly, according to Indeed.

Along those lines, a new report by online interview platform Karat found that the share of US engineering leaders now hiring for AI engineer roles (60%) has nearly doubled compared to last year (35%) – and they’re prioritizing AI skills:

  • AI engineering (74%)
  • Integrating AI functionality into products via API (62%)
  • Data science (58%)

The arrival of AI and genAI tools in the workplace has meant increased workloads and higher expectations for many workers. Both are major AI-related career issues: Indeed’s recent survey on genAI’s impact on workers found that 77% of AI users reported heavier workloads, yet nearly half were unsure how to use AI for efficiency.

Many of Indeed’s survey respondents said they face a lack of training, AI quality issues, ethical concerns, and disappointment with AI tools. That’s a critical mismatch with the 96% of C-suite leaders who expect AI to boost productivity, which is unlikely to happen until they align AI tools and training with workforce capabilities and expectations, the survey indicated.

Jobs and genAI

Indeed

Deloitte’s own survey found 68% of older workers — and 83% of early-career workers — now use AI. Tenured employees with AI experience remain just as likely to embrace AI when given the right learning opportunities. And that opens the door for internal training possibilities.

“An internal AI skills marketplace could connect them with projects to apply their skills and enhance strategic planning,” Lascaze said. “Formal mentorship programs, where early-career employees are paired with tenured staff, can become safe places for junior staff to experiment with problem-solving, learn how to navigate their careers and deepen their interpersonal skills.”

For December’s Patch Tuesday, 74 updates and a zero-day fix for Windows

Microsoft released 74 updates in its December Patch Tuesday update, with patches for Windows, Office and Edge — but none for Microsoft Exchange Server or SQL server. One zero-day (CVE-2024-49138) affecting how Windows desktops handle error logs requires a “Patch Now” warning, but the Office, Visual Studio and Edge patches can be added to your standard release schedule. There are also several revisions this month that require attention before deployment, including two (CVE-2023-36435 and CVE-2023-38171) that will need extensive testing. 

The Readiness team has provided this infographic outlining the risks associated with each of the updates this cycle. (More information about the previous six months of Patch Tuesday releases is available here.)

Known issues 

Other than the Roblox issue, Microsoft has published a reduced set of known issues for December:

  • There have been reports that the OpenSSH (Open Secure Shell) service fails to start, preventing SSH connections. The service fails with no detailed logging, and manual intervention is required to run the sshd.exe process. Microsoft has offered several mitigation options for those still affected.
  • For those still on Windows Server 2008 you might receive warnings that Windows Update failed to complete successfully. Microsoft is working on this issue and expects a fix to be released soon. Many users will now have to move to the second stage of “Extended Support Updates) or “ESU.”

Major revisions

For the final Patch Tuesday in 2024, there are these revisions to previously released updates:

  • CVE-2023-36435 and CVE-2023-38171: Microsoft QUIC Denial of Service Vulnerability. This is the third update to this two-year-old series of patches to the Microsoft .NET platform. Rather than a strictly information update, these patches will need to be added to your December release schedule.
  • CVE-2024-49112 : Windows Lightweight Directory Access Protocol (LDAP) Remote Code Execution Vulnerability. This is a release for this month’s update. This does not happen often, as this patch was only released 24 hours ago. (In fact, due to an error in the documentation, this patch was duplicated in the release notes as well.)
  • CVE-2023-44487: HTTP/2 Rapid Reset Attack. The update relates to a change in affected software — meaning all recent supported versions of Microsoft .NET and Visual Studio are included in the scope of the patch. Add this to your development update release schedule for the month.
  • CVE-2024-43451: NTLM Hash Disclosure Spoofing Vulnerability. This late edition revision has been widely reported in the news as it affects older versions of Windows Server (2008 and 2012) and has received some generous technical support from outside Microsoft.

This is an unusual month for revisions, with several patches from 2023 updated in the final months of 2024, with increased scopes and associated testing requirements. The Readiness team advises extra caution addressing both CVE-2023-36435 and CVE-2023-38171.

Windows lifecycle and enforcement updates

There were no product or security enforcements for this update cycle. However, Microsoft has noted that:

 “There won’t be a non-security preview release for the month of December 2024. There will be a monthly security release for December 2024. Normal monthly servicing for both security and non-security preview releases will resume in January 2025.”

Each month, we analyze the latest Patch Tuesday updates from Microsoft and provide detailed, actionable testing guidance based on a large application portfolio and a detailed analysis of the patches and their potential impact on the Windows platforms and application installations.

For this cycle, we have grouped the critical updates and required testing efforts into different functional areas including:

Networking and Remote Desktop Services

This month’s update addresses key components of Microsoft’s Remote Desktop Services with the following testing guidance:

  • Test RDP connections over the Microsoft Remote Desktop Gateway.
  • Try RPC over HTTP/HTTPS pathways while validating Remote Desktop broker features.
  • Test out DNS signing key operations for RRAS environments.
  • Validate WAN port operations (try netsh commands).

Local Windows File System and Storage

Minor changes to the Windows desktop file system will require a test of the ReFS system (light CRUD testing required). Due to changes in how Windows handles non-English characters, a test of Input Method Editors (IME’s) is required for Japanese formats. 

Virtual Machines and Microsoft Hyper-V

A minor update to a key virtualization driver will require some traffic testing and monitoring for Microsoft’s Hyper-V and virtualization platforms. While these recent updates are generally low-profile patches to Windows subsystems, we feel that the primary testing this month should focus on validating remote network traffic. The file system and Hyper-V changes require light testing. The goal for most enterprises is to get these Microsoft updates deployed before change control “lock-down” arrives.

Each month, we break down the update cycle into product families (as defined by Microsoft) with the following basic groupings: 

  • Browsers (Microsoft IE and Edge) 
  • Microsoft Windows (both desktop and server) 
  • Microsoft Office
  • Microsoft Exchange Server 
  • Microsoft development platforms (ASP.NET Core, .NET Core and Chakra Core)
  • Adobe (if you get this far) 

Browsers 

There were just two minor updates for Microsoft Edge this month, with CVE-2024-12053 and CVE-2024-49041 both rated as important. Add these low-profile changes to your standard release schedule.

Windows 

Though there is a strong focus on networking, this release also affects the following Windows features:

  • Windows Remote Desktop and related routing servers
  • Windows Kernel and Kernel Mode Drivers
  • Printing
  • Microsoft Hyper-V
  • Microsoft LDAP and LSASS
  • Windows Error Reporting

Unfortunately, there is a zero-day (CVE-2024-49138) that has been reported as publicly disclosed and exploited in the wild that affects how Windows creates error log files. Add these Windows updates your Patch Now cycle.

Microsoft Office 

Microsoft released nine patches to Office, all rated important. In addition, the company  offered some additional security measures and mitigations to the platform with the release of the advisory ADV240002, which covers the following areas:

  • Perimeter Defense
  • Network Security
  • Endpoint Protection
  • Application Security

This month’s update affects Microsoft Excel, SharePoint and core Microsoft Office libraries. Add these patches to your standard Office release schedule.

Microsoft SQL (nee Exchange) Server 

There were no updates for either Microsoft SQL or Exchange server.

Microsoft development platforms

Microsoft released a single update to the experimental AI music project Muzic with CVE-2024-49063. We’ll take this as a “win” with no further updates to Microsoft .NET or Visual Studio.

Adobe Reader (and other third-party updates)

Adobe has released a completely normal, run-of-the mill update to both Reader and Acrobat (Adobe Release notes). This is good news. This update has not been included in the Microsoft release cycle, which is as it should be. Adding to the huge, globally shared sense of relief, Adobe has chosen to modify its patching methodology to fall in line with industry best practices. Long-suffering IT admins have had to create (and maintain) “process workflow exceptions” to handle Adobe updates, usually with complex PowerShell scripts. No longer! 

Thank you, Adobe; there is no greater gift than a few less things to do (repeatedly).

For those readers who have enjoyed delving into the deeper details of all things patching, the Readiness team would like to say, “Thank you for the time and attention and we look forward to the New Year.” 

No surprises, right?

The Macy’s accounting disaster: CIOs, this could happen to you.

The Macy’s accounting nightmare is only getting worse, with the $24 billion retailer telling the SEC on Wednesday that both its annual report from last year and its auditor report “should no longer be relied on.”

Although the amount “hidden” was only $151 million — at the high end of Macy’s original estimate of “$132 million to $154 million” — the retailer said it exposed a massive weakness in its checks and balances procedures.

Macy’s did not get specific about the nature of the flaws, but the problem seems to be that the software charged with monitoring financial transactions was never designed to catch accountants doing what they do best: categorizing numbers in ways designed to make the company’s performance look better than it is. 

Such software is typically designed to catch true fraud, such as an employee exfiltrating money out of an enterprise into bank accounts they control, or payments to fraudulent contractors or even simple math errors. Apparently, the Macy’s system had weak safeguards that were easily sidestepped. Accounting officials say these same technology deficits likely exist in every enterprise. 

Macy’s “management identified a material weakness in its internal control over financial reporting related to the design of existing internal control activities involving manual journal entries over delivery expenses and certain other non-merchandise expenses, and the reconciliation of the related accrued liabilities,” the SEC filing said. “The Company identified that a single employee, who is no longer with the Company, intentionally made erroneous accounting entries and falsified underlying documentation, to understate delivery expenses from the fourth quarter of 2021 through the third quarter of 2024.”

When Macy’s first reported the incident, it used the word “hidden” and made no reference to “falsified underlying documentation.” Those are big clues about what likely happened. 

“The material weakness was the result of deficiencies in the design of controls over delivery expense and certain other non-merchandise expenses, and the related accrued liabilities, whereby the design of the controls did not consider the potential for employee circumvention of these controls,” the company said in its filing, adding there were “failures to obtain, or generate and use, relevant, quality information to support the functioning of these controls, including validation of the reliability of the information.”

Here’s the key “you’ve got to be kidding” point: “The design of the controls did not consider the potential for employee circumvention of these controls.” 

Really? The designers for an accounting system managing $24 billion in cash flow never considered that somebody might try to circumvent controls? Like perhaps someone engaged in naughtiness? 

The filing also showed some seeming contradictions. It stressed, for example, that this problem was done by just one employee — as though that’s a good thing. Imagine a Pentagon official explaining how 40 nuclear warheads were stolen and said, “I know this sounds bad, but this wasn’t done by a squadron on enemy fighters. This theft was just done by one guy, so all is fine.”

Macy’s also tried to say that this was not that big a deal. “The Company evaluated the errors and determined that the related impact was not material to results of operations or financial position for any historical annual or interim period.” 

But by the end of the filing, Macy’s attorneys used a lot of words to essentially say this actually was a big deal.

“As a result of the material weakness in the Company’s internal control over financial reporting described above, on December 10, 2024 the Audit Committee of the Board of the Company determined, based on the recommendation of management following its consultation with the Company’s independent registered public accounting firm KPMG LLP, that management’s report on internal control over financial reporting as of February 3, 2024…should no longer be relied upon. Additionally, KPMG LLP’s opinion as to the effectiveness of the Company’s internal control over financial reporting as of February 3, 2024 included within the Report of Independent Registered Public Accounting Firm in the Company’s Annual Report on Form 10-K for the fiscal year ended February 3, 2024, should no longer be relied upon.”

In accounting speak, declaring that their financials are not to be trusted is admitting that this is a big deal. Why? Given the lack of meaningful controls and strong safeguards in this one business unit, there is every reason to believe that the same lack of safeguards exist elsewhere in the company — and  according to accountants, in just about every enterprise.

Stefan van Duyvendijk, an industry principal with accounting software vendor FloQast, reviewed Macy’s filing and said that the retailer “is trying to distract people” by implying that the “small package delivery” unit is “the only place where Macy’s has this weakness.” 

This happened because that small package area was likely deemed low-risk, van Duyvendijk said, but Macy’s “reviews over journal entries are the same across the company.”

That means Macy’s likely knows that other similar issues could easily crop up — and that is what is tainting all of their reported financials and audits. 

The lone employee apparently reported that the small package unit owed less than it really did. “ERP is incapable of catching something like this,” van Duyvendijk said.

For other enterprises, this glaring hole in controls could be worse. The Macy’s problem appears— so far –to be one employee manipulating numbers to make the department look better.

It wasn’t outright fraud or theft. But that’s merely because the employee didn’t try to steal. But the same lax safeguards that allowed expense dollars to be underreported could have just as easily allowed actual theft.

“What will happen when someone actually has motivation to commit fraud? They could have just as easily kept the $150 million,” van Duyvendijk said. “They easily could have committed mass fraud without this company knowing. (Macy’s) people are not reviewing manual journals very carefully.”

Another accounting specialist,  JR Kunkle, an auditor and GRC specialist who runs his own consulting firm, Kunkle Consulting, agreed that the ERP and accounting systems used today can’t prevent accounting fraud in the way they should.

“If an individual is hellbent, he can change codes in the software. (Management) is going to rely on the accountant to setup the accruals,” Kunkle said. “Any kind of accounting entry requires judgment.” And today’s business software systems are incapable of reviewing and managing human judgment.

“Once you get inside (the accounting decision process) and there is a judgment factor, ERP can give you data about it, saying that it’s a shipping expense, but I don’t think systems in general can figure out what an accountant should enter,” Kunkle said. “I don’t know that you can automate that.”

Another financial specialist, Emburse CFO Andriana Carpenter, said that the software problem exists, but there areaccounting tactics that can minimize exposure.

“It’s true that most ERPs are not designed to catch erroneous accounting,” she said. “However, there are software tools that allow CFOs and CAOs to create more robust controls around accounting processes and to ensure the expenses get booked to the correct P&L designation. Initiating, approving, recording transactions, and reconciling balances are each steps that should be handled by a separate member of the team. There are software tools that can assist with this process, such as those that enable use of AI analytics to assess actual spend and compare that spend to your reported expenses. Some such tools use AI to look for overriding journal entries that reverse expense items and move those expenses to a balance sheet account.”

The specific problem Macy’s is struggling with could be minimized for others, she said. For example, someone bypassing safeguards can eventually be detected.

“In the event of management overriding accounting controls, leveraging the spend data on an end-to-end spend management platform and using AI analytics can identify this type of override by automatically comparing total spend to your P&L and identifying discrepancies,” Carpenter said. “In the case of this Macy’s accounting error, AI analytics would have identified differences in total payments versus the expense that was being reported.”

The ultimate problem here involves enterprise CIOs and their teams who trust software controls too much. Trusting software to religiously do what it is supposed to do is asking for trouble. Trusting that software to do what it was never designed to do? That is just demanding trouble.

Why would Apple make a Bluetooth chip?

Apple has already seized a leadership position with Apple Silicon. Now, it seeks to build a second bridgehead in networking chips so it can make the 5G chip, the Wi-Fi chip, and Bluetooth chips used inside its devices.

Why?

Cutting component costs might be part of its calculation, but improving performance, battery life, and the integration of these very different networking components might well yield a greater prize.

Apple now aims to introduce the first combined Wi-FI/Bluetooth chip in Apple TV and HomePod mini in early 2025, with the component set to appear in iPhones later next year. Macs and iPads will reportedly gain the new networking component in 2026. Current supplier Broadcom will continue to supply Apple with RF filters and is now working with it on development of AI chips for Apple’s Private Cloud Compute servers. (Broadcom and Apple also have some relationship on development of 5G modems for Apple’s devices.

To some extent, much of this was known. Apple’s silicon development teams have been working on multiple chips for use in Apple devices for some time, including M-, A-, S- and R- series chips used in Macs, iPhones, iPads, the Apple Watch and Vision Pro.

It also makes the W-series processors that manage Bluetooth and battery use on Apple Watch; the H-series chips (which are more efficient than W-family processors) inside AirPods; and the U-series family of UWB processors. There may be a handful of additional Apple-designed silicon components still in play in some older devices — it also made the T-series system management processors in late period Intel Macs. 

Apple is also a member of the Bluetooth Special Interest Group, which defines the Bluetooth standard. 

Apple’s big plan?

Development takes a lot of investment. Apple now has thousands of highly qualified engineering staffers working on silicon design for its fleet of devices. In Munich, Germany alone, the company now employs more than 2,000 people and we know it has others working on silicon development at key locations worldwide, including in the UK, where Apple CEO Tim Cook paid a visit this week. 

All this activity — and speculation Intel might try to poach Johny Srouji, Apple’s senior vice president for hardware technologies — represent the degree of investment Apple has been making in this sector.

Really, and truly, Apple has gone from zero to hero in processor design since it first invested in PA Semi. These investments mean the company now competes at the top of the silicon design industry and has the processors it needs to design and manufacture devices that just weren’t possible with other chips, opening doors for new types of hardware, wearables, and various forms of home/enterprise computing.

But while it isn’t clear how Apple can make a version of Wi-FI and Bluetooth that makes an inherent difference to its customers, the opportunities the project brings to product design seem a little clearer.

What benefits does this bring?

That’s the strategic benefit of what Apple has done so far in terms of core processor design (CPU, GPU, Neural Engine, etc). Now, Apple seems to want to achieve similar benefits in networking. What form could those benefits take?  There are some obvious possibilities:

  • Lower cost components: One way to keep retail costs stable is to control manufacturing costs.
  • Better integration: Think of it as the whole widget approach.
  • Energy use: Better heat dissipation and energy use should make for ever slimmer designs, potentially including the rumored iPhone 17 Slim.
  • Optimization: Better integration should make for improved networking stability.
  • Commodification: Apple gets to augment core networking features with Apple-only additions to benefit users.
  • Licensing: While unlikely, the company might want to license its networking technologies (even on a FRAND basis to improve its hand when negotiating other licenses). It may also want (or be forced) to put a “Made for Apple” licensing system in place to open up any Apple-only features to third-parties. 
  • Innovation: While Apple isn’t ready to do so yet, combining networking components on a single chip — or even eventually on a single SOC — should enable new opportunities, such as improved support for satellite communications. (Satellite is evidently part of Apple’s vision for networking, as will be HomeKit and biometrically controlled digital key deployments.)
  • Independence: Apple wants to reduce its reliance on third-party manufacturers for strategic components used in its devices.

Of course, these are just some of the possibilities. But to my mind, the biggest motivation will be to apply further differentiation to Apple’s hardware.

That’s not going to mean Apple will attempt to sell its devices on the merits of its own Bluetooth chip — that’s not Apple’s way. Its approach is to market its products on the basis of the features they bring. I think this means the integration of network services will form the foundation for new hardware features and services somewhere down the line, the most obvious being built-in LAN enhancements and satellite messaging.

On the latter, it is interesting just how many of the stories circulating in recent weeks seem to lead toward satellite, giving me a chance to grab my copy of Eddy Ramos’ Book of Laughs and say that when it comes to Apple’s future networking silicon adventures, you really should watch the skies. Unless you live in New Jersey or Oregon, where you may prefer to keep your eyes down to protect your sanity.

You can follow me on social media! Join me on BlueSky,  LinkedInMastodon, and MeWe

NotebookLM Plus is now available to Google Workspace customers

A new premium version of Google’s NotebookLM AI assistant is now available to Google Workspace customers.

Google unveiled NotebookLM last year, initially under the name Project Tailwind, and began testing the AI-powered “notebook” with select users. NotebookLM lets users upload multiple documents and other sources — Google Docs, PDFs, audio files and web URLs, for instance — that are analyzed by Google language models. Users can then query via a generative AI (genAI) chatbot interface. 

Google has added several features since NotebookLM was first unveiled, including  Audio Overviews, which generates a podcast-style audio discussion from the contents of uploaded documents. 

A free version of the app has been available to Google Workspace customers since September, and an early access pilot for an “enhanced” business version of NotebookLM was announced in October

Google also announced the launch of its new NotebookLM Plus. This is available to Google Workspace users that pay for the Gemini for Workspace add-on (which starts at $20 per user each month on top of Workspace subscriptions), as well as a standalone version via Google Cloud. Google didn’t immediately respond to a request for pricing for the standalone version of the app. 

The premium version removes some of the usage limitations with the free version. That means five times more Audio Overviews, queries, notebooks, and sources per notebook. There are also customization options for style and tone of user notebooks, and shared notebooks for teams with usage analytics.

Google highlighted “enterprise-grade” protections for business customers: employee uploads and queries entered into NotebookLM Plus won’t be used to train models and are not reviewed by humans, Google said.

“Your data remains your data and any files uploaded, queries and responses are not shared outside your organization’s trust boundary,” a Google spokesperson said in a blog post. 

NotebookLM users get access to a redesigned the user interface, too. 

“From the start, we wanted NotebookLM to be a tool that would let you move effortlessly from asking questions to reading your sources to capturing your own ideas,”  Steven Johnson, editorial director forGoogle Labs, said in a blog post. “Today, we’re rolling out a new design that makes it easier than ever to switch between those different activities in a single, unified interface.” 

The interface is organized into three components: a “sources” panel that manages information related to a user’s project; a “chat” panel, where you can query the NotebookLM chatbot about the contents of uploaded documents; and the “studio” panel, where new documents such as study guides, briefing docs and audio overviews, can be created with one click,Google said. Each component can be expanded and resized to help focus on a particular part of the app. 

There’s also a new feature under development: the ability for a user to interrupt an AI-generated Audio Overview conversation mid-flow and ask questions. “Using your voice, you can ask the hosts for more details or to explain a concept differently,” said Johnson. It’s like having a personal tutor or guide who listens attentively, and then responds directly, drawing from the knowledge in your sources.”

Google noted that NotebookLM will be embedded in Agentspace, a new tool for interacting with AI agents for work tasks. 

Google’s Agentspace will put AI agents in the hands of workers

Google has unveiled an AI agent builder tool designed to  automate repititive tasks and help workers find information held across their organization faster. 

AI agents have become a major focus for software vendors in recent months, including Atlassian, Microsoft, Salesforce, and numerous others. The “agent” concept is used in different ways, but generally refers to software systems that are able to take actions on behalf of a user, with varying degrees of autonomy. IDC analysts predict that at least 40% of Global 2000 businesses will use AI agents and agentic workflows to automate knowledge work, doubling productivity in the process — at least in cases where the technology is successfully implemented.

On Friday, Google unveiled Agentspace, its own application where workers can access and build agents. The standalone app  has three main purposes, according to Google.

One is to serve as the “launch point” for custom AI agents. These agents combine generative AI (genAI) large language models with multi-step workflows to automate repetitive tasks. Google said the application has an “intuitive interface” and intends it to serve as a space where workers can access pre-built agents created in Google’s VertexAI Agent Builder. A low-code tool is also in the works to enable a wider range of employees to set up their own agents.

Agentspace also provides an enterprise search function that Google said will help workers find information held in applications across their organization, includingboth structured and unstructured data such as documents and emails. Agentspace search is “multimodal,” Google said, meaning it should be possible to search across video and image files as well as text documents. 

Agentspace search can access data from a range of sources using connectors to third-party tools such as Confluence, Google Drive, Jira, Microsoft SharePoint, ServiceNow, and others. 

Users can interact with a conversational assistant that responds to search queries. Agentspace agents will also perform actions based on the information held in customers’ documents, Google said.

Finally, NotebookLM is also embedded in the Agentspace app. Unveiled as an “experimental” tool by Google Labs last year before a wider release in September, NotebookLM is billed as a “virtual research assistant” that provides responses grounded in documents and data supplied by a user. This includes the ability to create podcast-style voice summaries of selected documents, for example.

Agentspace is available now in early access with a 90-day free trial; it will require a monthly per user subscription fee after that period. Pricing details are yet to be announced, a Google spokesperson said. 

Google this week announced a range of AI “agent” tools, including  two research prototypes: Project Astra, which can perceive the physical world and provide assistance to users, and Project  Mariner, which understands and can take action on the contents of a computer screen. These are powered by Gemini 2.0, Google’s latest AI model which launched on Wednesday and is described by Google as its “model for the agentic era.”

18 indispensable Android travel apps

For all the ways travel’s evolved over the years, one thing has remained maddeningly steadfast: Moving from one place to another is almost always a hassle. There’s endless room for inconvenience and error, and a journey rarely goes according to plan.

But while there’s not much you can do about the late departures, the surly gate attendants, or the smelly fella somehow always seated right next to you, there are some tech-centric steps you can take to make your next business trip a little less unpleasant.

Android’s travel app selection has really taken off in recent years, and the Google Play Store now boasts an impressive array of genuinely useful titles for the traveling professional. After putting numerous standout candidates to the test, these are the apps I’d recommend stowing on your smartphone and keeping at arm’s reach whenever your work next has you hitting the road or flying the (allegedly) friendly skies.

(All apps are free unless otherwise specified.)

Android travel apps, part 1: Planning and preparing

Organize your packing process

PackPoint is a travel organization genie. You simply tell it where you’re going, when, and what you’ll be doing — and the app generates a detailed checklist of suggested items for your suitcase.

You can add your own items to the list, as needed, and then use it as a guide to make sure you remember everything, every time.

android travel app - packpoint

PackPoint takes some of the pain out of packing for a trip.

JR Raphael / IDG

PackPoint is free, with an optional one-time $3 upgrade that removes ads and gives you the ability to create your own custom packing templates. The paid version of the app also integrates with TripIt (more on that in a moment), which means it can import your travel plans automatically and create packing lists before you even ask.

Prepare for local navigation

Yeah, yeah, I know: You’re well aware of Google Maps. But what you might not realize — or maybe have just forgotten — is that with a teensy bit of planning, you can download all the data you need for a trip directly into Maps in advance. That way, you can navigate to your heart’s content, even in areas without strong mobile data signals, and you can avoid burning through mobile data unnecessarily on the road.

Here’s the trick: While you’re still in the comfort of your home or office, open up Maps on your phone and search for the city you’ll be visiting. Tap the city’s name within the search interface, then tap its name a second time when it appears in a panel at the bottom of the screen — or just swipe up on that panel to enlarge it.

From there, tap the More button in the row of options directly beneath the city’s name (and if you don’t see that button right away, try scrolling horizontally along that row to reveal it).

That’ll reveal a pop-up menu with an option to “Download offline map.” Tap that, then tap “Download” on the confirmation screen that appears. Once the download finishes, you’ll be able to access maps and directions within your destination without the need for an active connection.

Repeat as needed for any additional places on your agenda, then rest easy knowing your navigational guide will be there and waiting — no matter what sort of conditions you encounter.

Android travel apps, part 2: Flying

Manage your air travel

TripIt is an all-around air travel management companion, and it’ll make your life easier in some meaningful ways — especially if you do a fair amount of flying.

At its core, TripIt allows you to forward flight itineraries and other travel-related emails to a special address — or, if you want, to grant it direct access to your inbox so it can find and process such emails on its own — and it then extracts all the relevant details and organizes them into clean and easy-to-follow master itineraries.

Where TripIt really shines, though, is with its optional $49-a-year TripIt Pro service (which you can try out via a free 30-day trial). That service gives you real-time flight updates all throughout your trip — often beating notifications by airlines’ own apps, in my experience, as well as updates to the monitors in the terminal.

android travel app - tripit pro

Once you travel with TripIt — and specifically its TripIt Pro service — you won’t want to fly without it.

JR Raphael / IDG

Beyond that, TripIt Pro makes it dead simple to find alternate flights at any point in your adventure. If a connection is canceled or delayed, all it takes is a couple of taps to see what other flights are available — even down to the specific open seats — on your current airline or on another. That’s helped me stay a step ahead of the gate agent on multiple occasions when late departures have put connecting flights in jeopardy.

TripIt Pro comes with a few other perks, too, such as a two-month free trial of the CLEAR expedited airport access program. But the notifications and alternate flight finder are what really make the app invaluable. And while several other services offer similar sorts of travel planning features, no other app has been as consistently helpful, reliable, and easy to use as TripIt in my real-world travel testing. It’s the gold standard of travel organization and a must-have for any frequent flier or business traveler.

Find the best flights

Forget all the clunky, upsell-infested flight-finding services and instead, open up your Chrome Android browser and navigate to Google Flights. All right — so technically, it isn’t an Android app, but Google’s flight-searching system makes it super-easy to find and book flights across all airlines. You can save or share potential itineraries, monitor flights and get notified by email as soon as a specific fare goes up or down, and then buy your tickets directly with whatever airline (or airlines) you choose.

Pro tip: If you want to make the app easier to access, tap Chrome’s three-dot menu icon while viewing the website and select “Add to home screen.” That’ll give you a more traditional mobile-app-like icon that can then pull up the tool with a single tap.

One other utility that might be worth keeping handy is Hopper — but there’s a very specific purpose and also an important asterisk involved. Hopper watches flight prices over long periods of time in order to track trends and show you how fares are likely to fluctuate based on when you fly and when you make your purchase. If you’re booking your own travel and either footing the bill yourself or trying to stay within a limited company budget, that knowledge can be incredibly helpful to have.

android travel app - hopper

Hopper’s airfare-tracking system can give you valuable flight price knowledge.

JR Raphael / IDG

But Hopper’s ultimate goal is to get you to book your tickets through its service, and that doesn’t necessarily seem like the most advisable thing to do. User reviews on the Play Store mention difficulty changing itineraries once they’re booked with Hopper and challenges getting through to the company’s customer service.

So what I’d suggest is treating Hopper as a resource and not a ticket-purchasing portal: Use it to research optimal travel dates and purchasing windows, if you need to, and then take the info it gives you and plug it directly into either Google Flights or the appropriate airline’s website to buy the tickets directly from the source — and without the potentially problematic middleman.

Speed up your border entry

If you’re traveling internationally — and have a valid passport from the US or Canada, a US lawful permanent resident card, or a US Visa Waiver Program passport — the Mobile Passport Control app can save you precious time when you enter the US by letting you submit your passport info and customs declaration form ahead of time and then skip the regular line on your way through border patrol.

Despite what its name may suggest, though, the app doesn’t actually replace your passport. You’ll still need to carry that with you. It’s also currently supported only by certain airports, so you’ll want to make sure it’s available wherever you’re flying before you begin.

And not to worry: It’s created by the US Customs and Border Protections agency and 100% official and legit.

Android travel apps, part 3: Driving and public transit

Track your mileage

If you drive your own car for business, MileIQ — formerly owned by Microsoft but now back to being an independent entity — makes it as easy as can be to keep track of all your mileage for later reimbursement.

Once you set up the app on your phone and grant it the various permissions it requires, you don’t have to do a thing: It’ll just automatically detect when you’re driving and then log all your miles in the background. It even uses current IRS-mandated reimbursement rates to calculate what you’re owed.

The app has some interesting advanced options, too, such as the ability to set specific work hours and then ignore any drives that occur outside of those times.

MileIQ is free to use for up to 40 drives per month. For unlimited access, you’ll have to pony up $6 a month or $60 for a full year of service.

Pay less for gas

Why pay top dollar for top-offs when you can drive an extra minute from the highway and save yourself (or your company) some money? GasBuddy gives you the insight you need to find fuel that won’t break the bank: You just open up the app, tap the option to find gas near you, and then either look through a list of nearby gas stations and how much they’re currently charging or switch to a map view to see prices plotted out around your present location.

android travel app - gasbuddy

GasBuddy relies on user reports to provide up-to-date info on gas prices in your area.

JR Raphael / IDG

GasBuddy has a bunch of other features you probably won’t want to mess with, but the app’s price searching ability is worth every penny (particularly since the app is free and thus costs you precisely zero pennies to use).

Activate your highway X-ray

As anyone who’s ever taken a lengthy drive knows, fuel is only one tiny part of the highway exit decision matrix. Which exit you choose on your journey could determine if you end up with a gold mine of interesting options for dining, buying, and other delightful diversions — or if you find you’re facing a metaphorical (and maybe also literal) desert, with nothing of note anywhere around you.

An app called iExit will change the way you think about such choices.

iExit shows you a detailed breakdown of exactly what you’ll find at every exit on every interstate throughout the US, with a complete list of all restaurants, stores, parks, hotels, and other random attractions that are accessible from each exit you’re passing.

The app can use your current location to show you info as it becomes relevant, or you can manually search along any interstate to get an exit-by-exit overview. Either way, it’s completely free to use.

Find parking anywhere

When you’re traveling through a city, SpotHero will save you a substantial amount of time, money, and headaches with finding a place to park.

Just search the app for any specific location or let it scan your current location, and within a matter of seconds, you’ll see a list of available parking in the area — arranged by price, proximity, and even other drivers’ ratings.

android travel app - spothero

Searching for a parking spot is as simple as it gets with SpotHero by your side.

JR Raphael / IDG

You can often book a spot directly within the app, if you want — or you can just use it as a free and easy way to find a place to land.

Become a public transit master

If you’re ditching the car and relying on trains, buses, Ubers and Lyfts, or even bikes, scooters, and plain ol’ walking to get around your destination, Citymapper is the app you need.

In cities where it’s supported — a limited but reasonably extensive list — Citymapper lets you put in a starting and ending address and then explore the best ways to get from point A to point B using any combination of public transit options.

The app mixes and matches available methods and serves up an impressive menu of possibilities. You can pick the cheapest combo, the fastest path, or choose a specific way you prefer to travel and let Citymapper create a custom itinerary around that.

android travel app - citymapper

Citymapper goes above and beyond what Google Maps can do when it comes to public transit planning.

JR Raphael / IDG

Citymapper is free with an optional $1.50-a-month or $10-a-year ad-free upgrade.

Android travel apps, part 4: At your destination

Track your travel expenses

When it comes to more general expense-tracking, Expensify is the app to have in your arsenal. Expensify lets you simply take photos of receipts with your phone — or forward invoices and receipts via email — and it then extracts the relevant details and organizes them into reports. The app is available on the web as well, and it offers direct-export integrations with QuickBooks, Xero, and other accounting services.

android travel app - expensify

Snap a photo of a receipt — or forward it in via email — and then forget about it with Expensify.

JR Raphael / IDG

Expensify costs either $5 or $9 per person per month for businesses, depending on your needs. You can try the app out with a free individual plan, too, though that limits you to just 25 imports per month and lacks many of the service’s advanced expense reporting and integration options.

Find a place to stay on short notice

The next time you find yourself unexpectedly stuck somewhere — be it due to a cancelled flight or a road trip gone awry — don’t panic. Instead, snag the free and easy to use HotelTonight app. HotelTonight searches around your current location to find hotels with open and available rooms, but that’s not all: It also scores you legitimate savings on the rates, by way of an apparent deal wherein hotels let the service sell rooms at a discount in order to fill last-minute vacancies. I spot-checked a handful of the app’s recommendations, and the savings were absolutely real.

android travel app - hoteltonight

HotelTonight provides an easy way to find last-minute rooms at discounted rates.

JR Raphael / IDG

HotelTonight has handy details and ratings for all the hotels it recommends. And once you find something suitable, all it takes is a few taps within the app to book your room and be ready to roll.

Find Wi-Fi anywhere

Why waste money on mobile data when Wi-Fi is all around you and waiting for the taking? Just open WiFi Map to see an interactive map showing available Wi-Fi networks in your area (or any other area you want to search). The app lists out speed information and even provides user-submitted passwords to secured public networks in some instances.

Just note: When you first open WiFi Map, you’ll be pressed to upgrade to a $15-a-year premium subscription. That allows you to eliminate some rather aggressive ads within the app and also gives you the ability to download information in advance for offline viewing. You don’t have to make the upgrade, though (and arguably shouldn’t bother); if you want to use the app for free, just tap the little “x” in the upper-right corner of the screen when the upgrade prompt appears.

Convert and translate anything

For your next border-crossing journey, let XE Currency Converter convert currency for you without the usual headache. Once you tell the app your home country’s currency and select which foreign currencies you want to convert into, all you have to do is type in a dollar amount to get an instant glimpse at the exact equivalent based on up-to-the-minute conversion rates.

And when language translation is what you require, the aptly named Google Translate app is the tool you want. It’s jam-packed with practical features, such as the ability to translate text instantly from an image you capture with your camera and a “conversation mode” that lets you have a (somewhat awkward) back-and-forth dialog, in real time, with someone speaking a different tongue.

Stay fit wherever you go

Who says you have to stay sedentary just because you’re traveling? Skip the underwhelming hotel “exercise facility” and turn to AllTrails to find and navigate popular running, biking, and hiking trails wherever you are instead. The app is free to use, with an optional $36-a-year premium upgrade that gives you a variety of extra features like live sharing, offline-friendly downloading, and an ad-free experience.

(When you first start using the app, you’ll see a full-screen prompt to start a free trial subscription. Just note that you don’t have to do that and can skip over the offer altogether by tapping the “x” in the upper-left corner of the screen.)

If you’d rather get your heart pumping from the privacy of your own room, snag the Nike Training Club app. It’s filled with easy-to-follow workouts, ranging from the intense and Crossfit-reminiscent “Total Body Burnout” to the simple and stretch-oriented “Run Ready Flow.”

android travel app - nike training club

The Nike Training Club app has tons of workouts you can do almost anywhere.

JR Raphael / IDG

You can find workouts for practically any amount of time you want — as little as five minutes, even! — and you can browse specifically through “no-equipment workouts,” assuming you don’t carry your entire collection of kettlebells with you every time you travel. And best of all? The app is completely free to use.

The only thing you’ll be missing is an excuse.

This article was originally published in June 2018 and most recently updated in December 2024.