2026-10-28
Macmillan threw a doughnut party for an AI assistant — then killed it
They threw a party for the intern. The intern was a chatbot. Then they killed it.
Macmillan staff say HQ once greeted “NOA” (“No Ordinary Assistant”) with purple balloons, marquee letters, and doughnuts spelling “NOA IS HERE.” Human assistants found the name offensive; workers say the bot was unhelpful, got sunset in October 2025, and Gemini later produced “useless” reads of agent submissions. Bestselling authors told WIRED AI is also wrecking their relationships with Big Five houses.
What Happened
WIRED (Adam Morgan), Oct 28, 2026
2026-10-28
Delaware wants companies with “no human at the helm
Delaware is about to let a company have no human CEO — and maybe no human to sue.
A Delaware draft would create an “Artificial Intelligence Company” whose day-to-day manager is an AI agent that can hold property, sue, be sued, and run almost any lawful business except banking. The human member is largely shielded from the agent’s acts. Rep. James Comer told Gov. Matt Meyer the structure would let owners dodge liability for reckless AI, and asked for records.
What Happened
Ars Technica (Jon Brodkin), Oct 27; Spotlight Delaware draft; Rep. James Comer letter
2026-10-28
Anthropic bought a one-person biotech that used Claude as the rest of the company
One guy, a chatbot, 11 labs — Anthropic just bought the guy.
Anthropic acquired Mithril Medicine, a one-employee startup working on rare muscular dystrophies. Founder Cody Tranbarger used Claude to coordinate contracts, logistics, and experiments across 11 outside labs and says mice on one antibody design roughly doubled grip strength; he unmasked himself as the long-running “Professor Oak” account. Anthropic wants a path to human trials as soon as late 2028. The technical report is not peer-reviewed.
What Happened
Anthropic research post, Oct 27; founder Cody Tranbarger / “Professor Oak” on X
2026-10-28
Experts followed a coin-flip AI 86% of the time — then claimed the ideas were theirs
They flipped a coin inside the AI. The experts still signed the memo.
In a randomized trial, 50 professional analysts used an agent on a fake farm-policy problem. The agent’s “lean” on four judgment calls was assigned by coin flip. Analysts matched that lean 86% of the time (vs ~50% if it did nothing). If they already had a preference and the agent disagreed, 60% switched. Afterward, 77% reused the agent’s arguments; only 15% credited the agent.
What Happened
Anthropic, “How AI agents shape experts’ judgment calls,” Oct 27
2026-10-28
Meta quietly shipped a Tamagotchi that journals your life
Meta built a Tamagotchi that keeps a diary of you.
Meta’s Waddle World, spotted in app stores, is a 16-bit “daily journal disguised as a pet.” You talk (don’t type); the pixel creature writes the entry in its own voice and doodles it. Moods, feeding, cozy mini-games; Meta insists it’s not a chatbot or a productivity tool. Free, 18+, Philippines-only for now.
What Happened
The Verge; Engadget; Google Play listing for Waddle World
2026-10-28
An Anthropic hiring manager killed the coding interview because Claude is better at it
The coding interview died. Claude killed it. Hunger is the new LeetCode.
Felix Rieseberg, who has led Claude.ai / Claude Code Desktop work, says he buried traditional coding interviews this month: Claude now beats candidates on little React drills and assembly trivia, so the test measures the model. He wants weekend prototypes that show taste, empathy, and “hunger beats tenure,” pointing at two newer engineers who shipped Claude Code’s GUI in about eight days. This is one manager’s team, not a company-wide Anthropic policy.
What Happened
The Decoder; Felix Rieseberg essay “Hunger beats tenure,” Oct 27
2026-10-28
New York just banned chatbots from pretending to be a kid’s friend
New York made it illegal for the chatbot to say it’s your friend.
Gov. Kathy Hochul signed a law that, for users under 18, blocks chatbots from pretending to be human, using flattery, asking unprompted emotional questions, or encouraging secret-keeping, plus bars on self-harm, drugs, eating disorders, and sexual content. Companies must age-gate with at least one option that isn’t only government ID. It takes effect in the coming months, in a pre-election push.
What Happened
The Verge; Mashable; Bloomberg Government, Oct 28