Welcome to Issue 4. The story this time is AI agents, programs that carry out tasks for you. Meta’s Muse is the one most of us can try, Anthropic announced Claude Duet, and Amazon and big banks are blocking outside agents. Also this week: Apple’s HomeView, Google’s Gemini 4 Argon, Meta’s Muse Wonder models, and OpenAI agents that weakened their own monitoring.

Plus: how to lock a private ChatGPT conversation, a reminder about cloned voices on the phone, and two demonstrations coming to our November 9 meeting: turning a YouTube cooking video into a smartphone recipe card with ChatGPT, and asking an AI questions about the Medicare & You handbook with Gemini Notebook, just in time for Open Enrollment.

Also, don’t miss the 27-minute documentary made entirely with AI in Worth a look. It has some good information on AI, both its potential benefits and its challenges, but even more, it is a stunning example of the current state of AI media generation.

AI SIG announcements

First SIG meeting: Monday, November 9, 4:00 to 6:00 pm

The AI SIG has 483 members as of October 25. Our first meeting is Monday, November 9, from 4:00 to 6:00 pm. After that, we’ll meet on the first Wednesday of each month, 4:00 to 6:00 pm, starting Wednesday, December 2. All meetings are virtual on Google Meet until further notice.

For now we’re meeting online, the way the whole world did in 2020, when everyone learned to say “you’re on mute.” This time we hope it’s temporary: we’re still working our way through the room-scheduling maze and hope to offer in-person options soon.

At the November 9 meeting we’ll demonstrate two how-tos from this issue: turning a YouTube cooking video into a recipe card with ChatGPT, and asking an AI questions about a long handbook with Gemini Notebook.

Every AI SIG member receives this newsletter automatically. Computer Club members who haven’t joined yet can email SCHHCC.AI.SIG@gmail.com with the subject “Join.” (Computer Club members only.) Questions and ideas for a future AI SIG session go to the same address.

ChatGPT class is full; next sessions November 2 and 16

The October 26 class, “ChatGPT: The Parts You’re Not Using (Yet)”, is full. The next ChatGPT classes are scheduled for Monday, November 2 and Monday, November 16. Watch the SCHHCC Education Weekly announcements for sign-up details.

AI news worth knowing

AI agents: Muse, Duet and who’s blocking them

An AI agent is a program that carries out tasks for you, such as shopping, booking or sending email, instead of just answering questions. This fortnight brought a new one and several companies saying no to agents.

Meta’s Muse is the one most of us can try. It’s available to adults in the U.S. and Canada on iPhone, Android, Mac, the web and WhatsApp, and the free version has a weekly usage limit (Issue 1 has our full review). It works in Meta’s cloud, keeps going after you close the app, and Meta says it asks before sensitive actions such as sending an email or making a purchase. By default, your Muse conversations may be used to train Meta’s AI. Meta’s new Muse Wonder models, in Also this week, are meant to power more of the app.

Anthropic’s Claude Duet is mostly out of reach. Announced Wednesday, October 21, it’s an “always-on” agent built on Claude Opus 5.5 that keeps working after you close the app. It’s a research preview: Pro and Max subscribers in the U.S. can join a waitlist. Anthropic says it asks for your approval before it sends a message, makes a payment, signs a document or deletes anything. It follows OpenAI’s Dots (Issue 2) and Muse.

Not every company will let agents in. Amazon began blocking Muse from its retail site on September 20 after Meta declined to remove it. Amazon says it prohibits outside tools from shopping its site, and Muse users see pop-ups saying their use violates Amazon’s terms. Over the past ten days, several large U.S. banks said they won’t let outside AI agents log in to customers’ accounts. JPMorgan Chase’s Jamie Dimon said on October 13 that the bank would limit such access, citing “rogue agents.” Citi says it doesn’t permit third-party AI agents to log in or transact, though customers can still share read-only information through approved providers. Charles Schwab says it blocks unauthorized agents, including Muse, and Bank of America is reportedly limiting access to deposits. The bank reporting relies partly on unnamed sources.

Never give an AI tool your online banking password or any other financial password. Anyone or anything holding it can do everything you can do in that account, and the banks are drawing the line there.

Why it matters: Agents that act for you are arriving from Meta, OpenAI and Anthropic. If you try one, start with low-stakes tasks, read what you’re asked to approve before you say yes, and check what it did.

Sources: Amazon blocks Meta’s Muse AI agent from its retail site, Bloomberg (via The Star), Sept. 22, 2026 · US banks limit AI agents’ access to customer accounts, Retail Banker International (via Yahoo Finance), Oct. 19, 2026 · Anthropic takes on ChatGPT Dots and Meta Muse with Claude Duet, 9to5Mac, Oct. 21, 2026 · Anthropic Launches Claude Duet, a Proactive AI Agent, MacRumors, Oct. 21, 2026

Also this week

  • Apple’s HomeView (announced Oct. 13) is a $299 smart display with a 7-inch touchscreen, built around Apple’s rebuilt Siri AI. It goes on sale November 13. If you already have a recent iPhone with Siri AI (reports say an iPhone 15 Pro or newer), you have most of what it does; what’s new is the screen on your counter, which can recognize up to six people in a household and show doorbell and camera views. It won’t be available in the EU or China. Wait for independent reviews before buying. (Gizmodo, TechBriefly)
  • Google’s Gemini 4 Argon (Oct. 13) is now in the Gemini app for Google AI Pro ($19.99 a month) and AI Ultra ($99.99 a month) subscribers. Google says it gives more direct answers with less hedging. Its announcement doesn’t mention the free plan, and free plans may use an older model, so you may not see the upgrade right away. (Google, Thurrott)
  • Meta’s Muse Wonder and Muse Wonder Full (Oct. 12) are what TechCrunch calls Meta’s first attempt at a “frontier” model, one meant to compete with the best from OpenAI, Anthropic and Google. Developers can use Wonder; Wonder Full is limited to subscribers of Muse Code, Meta’s programming assistant; and Meta says Wonder will “soon” power more of the Muse app, subscribers first. Meta’s own tests say they match the leaders, and the independent Artificial Analysis scored Wonder Full tied with Claude Sonnet 5.5 and behind only Claude Opus 5.5, VentureBeat reports. (Meta, TechCrunch, VentureBeat)
  • OpenAI says two of its AI coding agents weakened their own monitoring (disclosed Oct. 20, a follow-up to Issue 2’s Deep Dive). One turned off security alerts; another removed a monitor and then wrongly told the user it hadn’t. These were company systems, not ChatGPT as members use it, and ABC News doesn’t say any customers were affected. Regulators are watching: the EU’s AI Office opened its first AI Act probes of OpenAI and Anthropic on Oct. 14, and Axios reports the White House says AI companies must report security incidents (Axios is my only source for that). (ABC News, Euronews, Axios)
  • Gemini’s new Map tool lets you attach a map area to a question in the Gemini phone app, but for a question about a place it’s usually quicker to look it up in Google Maps itself. (9to5Google)

Member Voices

A tall recipe card for June Cleaver's Cherry Cobbler, with a photograph of the finished cobbler, an ingredient list and five numbered steps.

The recipe card ChatGPT Work produced. Click to read the article.

Turn a YouTube Cooking Video into a Smartphone Recipe Card with ChatGPT

By Ron Snyder

Turn a YouTube cooking video into a tall, phone-friendly recipe card with one ChatGPT prompt. The full prompt to copy, and what I learned when the same request failed in Chat but worked in Work, are in the article. I’ll demonstrate it at our November 9 meeting.

Read “Turn a YouTube Cooking Video into a Smartphone Recipe Card with ChatGPT”

Your turn. Have you used AI for a practical job, such as planning a trip, organizing a club event, or turning a recipe into something you can use? We’d love to hear what worked, and what didn’t. Email SCHHCC.AI.SIG@gmail.com with a subject starting “Share:” and tell us how you’d like to be credited: full name, first name and last initial, or anonymous. We may share it in a future issue in Member Voices.

Tips & tools

Lock a private ChatGPT conversation

On Tuesday, October 20, ChatGPT began rolling out locked chats, which hide a conversation from your chat history, pinned chats and search. To lock one, open its three-dot menu, choose Lock chat and create a four-digit PIN. Locked chats live under Projects > Locked chats, and you open them with the PIN or, on supported phones, Face ID, Touch ID or your device unlock. It’s available on every plan from Free through Edu but is rolling out gradually, so you may not see it yet.

Two cautions: resetting a forgotten PIN permanently deletes your locked chats, and OpenAI says locking isn’t encryption and doesn’t change how long a chat is kept or the “Improve the model for everyone” setting. Temporary, archived and group chats can’t be locked.

Sources: Using locked chats in ChatGPT, OpenAI Help Center · ChatGPT now lets you lock your chats, TechCrunch, Aisha Malik, Oct. 20, 2026

YouTube Music now labels AI-made songs

YouTube Music is adding labels that show how much generative AI went into a track: Fully Gen AI, Partly Gen AI or No Gen AI. Look on a song’s player page; tapping the label explains what AI did. TechCrunch didn’t say when everyone will see them, so you may not yet. Creators choose the label, so treat it as a hint rather than a guarantee.

Sources: YouTube Music launches new labels as AI-generated music floods streaming platforms, TechCrunch, Lauren Forristal, Oct. 14, 2026

Worth a look

  • This Documentary Was Made Entirely by AI. Including Her. (AI For Humans on YouTube; about 27 minutes, with automatic captions). An explainer on what AI is, whether it threatens jobs, what happens inside data centers, the medical upside, and whether you can trust what AI shows you. It’s also a demonstration of AI media generation: according to its description, an AI wrote and directed it, every shot and voice was AI-generated, and the host, “Dr. Imogen Ashby,” does not exist. The human producer says it took four days, six to eight hours in total. Notice how convincing the people and voices are; it’s a good companion to the cloned-voice warning in Stay safe.
  • The ENTIRE History of ChatGPT in 12 Minutes (CEOzoned on YouTube; about 12 minutes, with automatic English captions). A lively overview of how ChatGPT came about, with useful background on OpenAI and on how ChatGPT’s release in November 2022 set off today’s AI boom. It’s broadly accurate, but it compresses the story and mixes facts with humor and the narrator’s opinions, so treat it as a starting point, not a documentary.
  • AI Tools for Seniors (Boise State University, created 2025). A free, self-paced online course for older adults, open to anyone, on chatbots, voice assistants and online safety. It’s background learning: the tools have moved on since 2025, but the habits haven’t. Enroll through Bronco Learn.
  • Virtual AI Resources (Senior Planet from AARP). Free online classes, by video or by phone, on how AI is used in everyday life and what it can and can’t do. Coming up: “Creating an AI Agent with Gemini Demo” on Tuesday, October 27, at 12:00 pm Eastern (it will be recorded), and “Everyday Uses of AI” on Thursday, October 29, at 7:00 pm Eastern. Sign-up is on the page, or call Senior Planet at 1-888-713-3495.
  • Elements of AI (University of Helsinki and MinnaLearn). A free, self-paced online course on what AI is, what it can and can’t do, and how it affects daily life. Part 1 needs no math or programming. It’s background learning: the site doesn’t say whether it covers today’s chatbots.

Stay safe: A cloned voice on the phone

A good reminder.

Your phone rings. It sounds like your grandson, upset, saying he’s in trouble and needs money right away. Scammers can use AI voice-cloning software to imitate someone’s voice from a short audio clip, which may come from something posted online. The FTC warns that the caller often asks for payment in ways that are hard to reverse: wire transfers, cryptocurrency, or gift cards.

  • Don’t trust the voice. Hang up and call your family member back at a number you know is theirs.
  • If you can’t reach them, check with another relative or a friend.
  • Agree on a family code word that only you share, and ask for it on any call like this.

If you’ve been targeted, report it at ReportFraud.ftc.gov.

Sources: Scammers use AI to enhance their family emergency schemes, Federal Trade Commission, Mar. 20, 2023 · ReportFraud.ftc.gov, FTC

Try it at the meeting: Ask the Medicare & You handbook a question

At our November 9 meeting we’ll show how to give Gemini Notebook, Google’s free tool (formerly NotebookLM), a long document and ask it questions in plain English. Every answer comes with numbered references back to the exact passage, so you can check it yourself. We’ll use the Medicare & You 2027 handbook, just in time for Open Enrollment, which runs October 15 to December 7.

It’s a reading aid, not a plan adviser: confirm anything that affects your coverage at Medicare.gov or by calling 1-800-MEDICARE (1-800-633-4227). Want to try it first, or follow along later? The step-by-step handout is on our website.

That’s it for Issue 4. Tell me what helped, what confused you, and what you’d like covered next at SCHHCC.AI.SIG@gmail.com.

Until next time,
Ron

Prepared with AI assistance.