← All issues
Issue No. 10

10 trail markers

Privacy locks in ChatGPT, a carefully qualified medical study, Gemini on the road and in spreadsheets, a new Texas lawsuit, the data-center power proposal, two robot announcements, an AI-use survey, and INTERPOL’s new campaign — plus practical and campfire reading.

Ten trail markers this Sunday, with a little more room below for trying something useful. Related stories are labeled as updates; research results and company promises are not the same as proven benefits.

From the Publisher: A little more room for the useful part

Howdy from Fort Worth. This week I’m trying a little more room for the useful part: a short news roundup, followed by something you can put to work and a little food for thought.

I’m less interested in which machine wins the latest contest than in what helps us stay independent, ask better questions, and avoid an expensive mistake. The extra sections are an experiment, not a new commitment. Tell me what earns its place and what you would leave out.

01
Privacy & Practical AI

ChatGPT adds a lock — but a PIN is not a privacy policy

OpenAI’s October 20 release notes introduce Locked Chats: selected conversations can be put behind a four-digit PIN, with availability rolling out gradually. That may help keep a personal conversation out of casual view on a shared computer. It does not change the service’s data-retention or model-improvement settings.

Check whether the feature has reached your account before relying on it. Read the reset warning carefully: resetting the PIN permanently deletes locked content. Locking a conversation also does not erase details the assistant previously remembered. My take: useful protection from a curious glance, not permission to upload tax returns, medical records, passwords, or account numbers.

Source: OpenAI · Locked Chats
02
Health & Wellness

An AI scores well on medical puzzles; the real-world test comes next

TechCrunch reported October 22 that Microsoft’s MAI-DxO system reached the correct diagnosis in 83% of simulated complex medical cases, versus 36% for individual primary-care doctors and general internists using reference materials. Specialists and physician teams were not the comparison group. Microsoft plans an evaluation with Beth Israel Deaconess Medical Center in Boston.

This is a new study, not the earlier story about patients asking a chatbot for clues. A difficult-case score does not establish that AI is better than your doctor in everyday care, or that it improves patient outcomes. The promising direction is a second set of eyes for clinicians. Ask who reviews any AI recommendation; don’t change treatment on a chatbot’s say-so.

Source: TechCrunch · October 22📎 Source: Microsoft AI
03
Everyday AI

Gemini can look up your appointment while it rides along

Google began rolling out Personal Intelligence for Gemini on Android Auto on October 20. With permission, it can use connected information such as Gmail, Calendar, and Maps to answer a request like “Take me to my doctor’s appointment.” Cars with Google built-in are scheduled to receive it in the coming months, not all at once this week.

It is opt-in, but the preference can carry over if you already enabled it in the Gemini phone app. Availability depends on an eligible personal Google account; business and education accounts are not included. Check permissions and test it while parked. Ordinary navigation, music, and calls still work without personalization. Convenience should not cost you attention to the road.

Source: Google · October 20📎 Source: 9to5Google
04
Practical AI

Google Sheets on iPhone gets a helper for the numbers

Google announced October 19 that Gemini in Sheets is reaching iPhones and iPads. Eligible subscribers can ask questions about a spreadsheet, request a summary, or create a chart. The gradual rollout began October 15 and can take up to 15 days. Eligible Google AI and Workspace plans are required; this is not a promise to every free account.

Try it with a club expense sheet or a small, nonprivate household example: “Summarize this table and flag any totals that need checking.” Compare the result with the original cells and your calculator. It is useful for finding a pattern, not an excuse to stop checking arithmetic. You don’t need to buy a subscription just to keep up.

Source: Google Workspace Updates
05
The Big Roundup · Texas Update

Texas files its own case against OpenAI

Last week we covered eight states’ lawsuits. This is a separate development: on October 23, Texas Attorney General Ken Paxton filed a case in Collin County against OpenAI and Sam Altman, alleging ChatGPT harmed Texans and lacked adequate safeguards. The state seeks restrictions including blocking minors’ access and court-appointed review before new models are released.

These are allegations and requests, not court findings or rules already in force. The Texas Tribune reported that OpenAI did not immediately respond to its request for comment. For context, the Tribune also notes that Paxton is running for the U.S. Senate and that Elon Musk, whose Grok competes with ChatGPT, is funding his campaign. That context does not settle the allegations. The new point for readers is the Texas filing; the family-safety concerns themselves are not new.

Source: The Texas Tribune via KBTX
06
Energy & Data Centers · Policy Update

Trump calls for data centers to bring their own power

In an October 21 Truth Social post quoted by Fox News, President Trump said new, large data centers should generate their own electricity and give or sell excess power to local communities or the grid. He also urged companies to offer community “dividends.” This is a new proposal in the power-cost story we have been following.

The cited report does not establish the details of an enforceable requirement, a payment program, or a reduction in your electric bill. Texans should watch the actual agreements: who pays for new equipment, who carries the risk, and what commitments are binding? A presidential statement is a signal to follow, not a household rebate to count on.

Source: Fox News · October 21 coverage
07
Robots · Development Update

Figure’s next robot passes a design milestone — not a home-readiness test

In a company presentation published October 22, Figure CEO Brett Adcock said Figure 04 had reached “design lock,” with parts shipping and component testing continuing. Humanoids Daily reported the update October 23. The announcement does not give a household price or a date when readers can buy one for home care.

Design lock is a development milestone, not proof the robot is finished, safe, or dependable. Battery, hand, and actuator work continues. This is worth a brief look because it is a distinct hardware update, not another promise that robots will soon do everything. Before any future purchase, ask for independent testing of the exact chores you need.

Source: Humanoids Daily · company presentation
08
Robots at Home · Company Announcement

MOBO is pitched as a companion, not a caregiver

AiMOGA’s company-issued release, distributed October 19, announced plans to unveil MOBO on October 21. The company describes voice interaction, object recognition, schedule reminders, and conversational memory. That release establishes an announcement, not independent confirmation of a completed launch or tested household performance.

I’m keeping this as a short company-announcement item, not a buying recommendation. Talking and remembering do not demonstrate safe caregiving. Before trusting any such device, ask about availability, recurring costs, data deletion, who can access recordings, and what happens when the connection fails. A reminder device should support a care plan, not replace a person.

Source: AiMOGA · company release via GlobeNewswire
09
Reading the News

More than half of surveyed voters use AI weekly — not “half of seniors”

Navigator Research published an October 20 report finding that 56% of surveyed registered voters said they used AI at least weekly. The online survey was conducted October 1–5, with a main sample of 1,000 registered voters and additional subgroup interviews. It is self-reported use, not a count of everyone who opens an AI app.

The number does not tell us that 56% of older adults use AI, or that you are behind if you don’t. My take: learn one useful task at your own pace. A survey can describe a trend; it cannot decide which tool is worth your time or money. The same report finds people see both promise and risk — a sensible combination.

Source: Navigator Research · report and methodology
10
Watch Your Back · New Campaign

INTERPOL takes the deepfake warning to an international audience

INTERPOL launched a public-awareness campaign October 19 about AI deepfake scams, including criminals impersonating relatives and other trusted people through synthetic voices or faces. The new development is the campaign itself. We have already covered family-emergency voice scams, so I’m not passing the same warning off as a new discovery.

The short reminder: a convincing voice or face is not verification. Independently reach the person on a number you already trust before sending money. A family question or code word is an extra safeguard, not absolute proof. If money has already gone out, contact your bank or payment provider promptly. The campaign is new; the need to slow down and check is familiar.

Source: INTERPOL · October 19 announcement

In the Saddlebag: Ten minutes to organize a busy week

A useful AI task needn’t be dramatic. Start with a few ordinary chores: a library visit, grocery shopping, a club meeting, and a call to a friend. Leave out addresses, account numbers, and private family details.

Try this: “Organize these tasks into a simple weekly checklist. Use only the details I give you. Don’t invent appointment times. Put anything with a deadline first, leave room for breaks, and mark missing information as ‘confirm.’ Here are my tasks: [your list].”

Read the result against your real calendar. Then ask: “Make it shorter, use larger headings, and leave a blank space for notes.” Print it or copy it into your own calendar. Nothing gets booked or sent simply because a chatbot has written a plan.

The point is not to hand over your week. It’s to spend less time sorting it out.

Healthy Aging: Let your care team set the limits

AI can turn instructions you already have into a clearer checklist. It should not decide what exercise is safe for your balance, change a medicine, or set a treatment plan.

If your clinician has given you an activity plan, try: “Turn these instructions into an easy-to-read weekly checklist. Keep the activities and limits exactly as written. Do not add exercises or change the amounts. Flag anything unclear for me to ask my clinician.”

Remove identifying details first, and compare every line with the original. If something is missing or different, keep the clinician’s instructions. You don’t need an AI-designed workout to make good use of AI.

Around the Campfire: A word game, not a health test

Here’s a change of pace with your Sunday coffee. Read these ten words, cover the list, do another small task, and then write down the words you remember.

Horse · River · Lantern · Apple · Garden · Clock · Piano · Bridge · Coffee · Mountain

For another round, ask AI: “Give me ten different everyday words for a relaxed memory game. No scoring labels or medical conclusions.”

This is a game, not a screening test. A score cannot diagnose or rule out a memory problem, and one puzzle does not establish a health benefit. If a change in memory is affecting daily life, talk with your healthcare professional.

Cowboy Term of the Week: Humanoid Robot

A humanoid robot has a roughly human-shaped body, often with arms, hands, and legs. That shape can help it use spaces and objects built for people. It doesn’t tell you how capable, independent, or safe the machine is. A convincing demonstration is not the same as dependable help in your home.

Until Next Sunday

What everyday task would you most like a little help with? And does this longer roundup feel useful, or would you rather keep it short? Send your thoughts to dward@ddub.ai.

Happy trails from Fort Worth,
Dennis “D Dub” Ward
The AI Cowboy · ddub.ai