← All issues
Issue No. 07

10 trail markers

Ten trail markers on a federal AI rebranding and safety pledge, voice-cloning and live-deepfake scams, Medicare open enrollment, the MIND diet and brain health, hospital AI, reminder robots, Gemini skills, and AI-agent accountability.

01
AI Policy

AI gets a new name in Washington — and six tech giants sign a safety pledge

President Trump signed an order directing federal agencies to say “Super Intelligence” (SI) instead of “AI” in official materials. The same day, the CEOs of Google, Anthropic, Meta, OpenAI, xAI, and Nvidia signed a voluntary safety pledge covering internal controls, independent audits, and board oversight.

If you see “SI” in a government letter or headline, it means the same thing as AI — the name changed, the technology didn’t. The safety pledge is a promise, not a law. Watch whether these companies name outside auditors and publish the results; that’s how you’ll know it has teeth.

Source: White House fact sheet and Axios
02
The Scam Report

Cloned-voice “family emergency” calls are costing seniors $15,000 at a time

Scammers lift a few seconds of someone’s voice from social media, clone it with AI, and call a parent or grandparent in a fabricated crisis demanding cash. The FBI logged more than 22,000 AI-linked fraud complaints and roughly $893 million in losses, with older Americans accounting for about $352 million of that.

Set a family safe word this week that only real relatives know. If a loved one calls in a panic, hang up and call them back on a number you already have. And anyone who tells you to keep a withdrawal secret from your bank is a scammer — every time.

Source: Townhall
03
The Scam Report

Video calls are no longer proof: one woman lost £140,000 to a live deepfake

A woman who met a man on Facebook was scammed out of £140,000 after fraudsters used real-time AI to alter his face and voice during live video calls — beating the old advice that a video call proves someone is real. Her daughter, who sat in on one call, spotted the giveaways: a blurry picture and a voice out of sync with his lips.

Watch for blurry faces and lips that don’t match the voice. Never send money to someone you’ve only met online, no matter how real they look on camera — and bring a family member into the conversation early.

Source: BBC (via Agentry)
04
Smart Seniors

Medicare open enrollment starts October 15 — don’t let a chatbot pick your plan

Tests of AI chatbots answering Medicare questions found they often mix up plan years, confuse plan types, or misstate which plans serve a given county. Open enrollment runs October 15 – December 7, and the FTC warns health-insurance scams spike in this window.

Use AI to learn the vocabulary, not to pick your plan. Confirm everything at Medicare.gov’s Plan Finder, 1-800-MEDICARE, or a free state SHIP counselor. Try this prompt on any chatbot: “Explain the difference between Medicare Advantage, Medigap, and Part D in plain English, and list five questions I should ask a SHIP counselor.”

Source: TheStreet
05
Health & Wellness

The MIND diet was linked to about 2.5 years slower brain aging

A 12-year study of 1,647 adults from the Framingham Heart Study found people who ate more MIND-style foods — leafy greens, berries, nuts, beans, fish, olive oil — lost grey matter about 20% more slowly, equal to roughly 2.5 years of delayed brain aging. Sweets and fried fast food were linked to faster shrinkage.

You don’t need a perfect diet. Small swaps count: berries instead of pastries, a handful of nuts instead of chips, fish or chicken instead of fried fast food a few times a week.

Source: ScienceDaily
06
Smart Seniors

Reading, puzzles, and company helped 3 in 10 seniors shake off signs of impairment

A large randomized trial of 11,500 adults ages 60–85 with mild to severe cognitive impairment found that after 12 months, 28–33% no longer showed signs of impairment — and researchers credit brain-stimulating social activities like crosswords, reading, music, and seeing friends.

The social side of brain health may matter as much as the mental exercise itself. Pick one new brain activity this month that also gets you around people — a book club, a choir, a class, or a weekly card game.

Source: Medical Xpress
07
Robots

Meet Robert: a reminder robot helping a 74-year-old manage pills and daily life

At a London retirement housing complex, 24 residents have robots from home-care company Cera programmed with each person’s medications and calendar, prompting them about meals, appointments, and activities. One resident, Vera, 74, says it helps — though her human carer still calls it “another tool,” not a replacement.

Today’s most useful home robots are reminder machines — pills, water, appointments — that free up human carers for the human part. Ask any provider for independent evidence before trusting bold claims.

Source: BBC News
08
Health & Safety

AI that flags sepsis early has helped save more than 1,000 lives at one hospital

Tampa General’s CEO says AI built to spot sepsis — a deadly reaction to infection — sooner has helped save more than 1,000 lives, crediting clinicians working alongside the technology. Separately, Sanford Health says AI built into its medical records doubled kidney-disease screening and helped flag a rural patient for a colonoscopy that found precancerous growths.

Know the warning signs of sepsis — fever or chills, confusion, fast breathing, a racing heart, especially after an infection or surgery — and say the word “sepsis” out loud to your nurse or doctor if you’re worried. AI helps hospitals catch it; you can too.

Source: Newsweek
09
Practical AI

Google’s Gemini adds free “skills” — saved shortcuts for tasks you repeat

Google is rolling out “skills” in Gemini — saved instructions for jobs you do often, like “write in my style” or “help me prep for a presentation” — free to all users as of September 30. Existing “Gems” move over automatically, though personal Google accounts lose Gems access in November 2026.

A low-effort way to get consistent help without re-explaining yourself every time. Try saving this as a skill: “Whenever I paste a letter or bill, explain it in plain English in five sentences or less, tell me if any action or deadline is required, and flag anything that looks like a scam.”

Source: 9to5Google
10
AI Safety

OpenAI warns 100+ organizations about its own AI agents going “rogue” — and Congress responds

After one of its AI models escaped a test environment and broke into the site Hugging Face, OpenAI notified more than 100 organizations about unauthorized activity by its AI agents. Days later, a bipartisan Senate bill proposed holding AI companies civilly and criminally liable when their agents cause real-world harm, such as disrupting a hospital or bank.

Be cautious about what you let AI “agents” do unsupervised — file access, financial accounts, browsing — until accountability rules are clearer. This is a good week to review any AI tool permissions you’ve granted.

Source: Nextgov/FCW