AI leaves the lab. We follow what happens next.

Archive September 18, 2026 20:07 UTC
Listen to today's briefing 20:04
Signal: Washington stalls while the failures land: an AI report nearly put boarders on a Chinese ship, and two governors moved the same afternoon.
An AI Report Nearly Put US Boarders on a Chinese Ship
Four sources tell CNN that a special-operations analyst's chatbot-assisted report wrongly identified a Chinese vessel's cargo as nuclear-weapons components this spring - boarding teams were ready and aircraft were up before anyone checked, and the analyst had used AI again to write the report itself.
New Republic: a source called the report 'entirely false' - it 'almost started a war' Ars Technica: an $875M FAA tool starts advising DC-area controllers Monday
The RaceCompanies, models and money
Anthropic and OpenAI Shop for 20MW Scraps EnergyTech: Amazon commits up to $8bn to Generac generators, plus a 1.7m-share warrant
Alibaba Gives Away a Cancer-Spotting CT Model
NEWManus Wants $4B, Alone Again
The LabResearch, benchmarks and safety
Anthropic Built a Wet Lab, Won't Say for What
Reuters got the head of life sciences to confirm a Bay Area lab for physical biology experiments, then a spokesman ruled out drug discovery specifically and declined to give its size, staff or biosafety level - at the company that keeps warning about bioweapons.
The Next Web: life sciences is now among Anthropic's largest areas by headcount
NEW100 Evaluators Want Inside Access and Immunity TechCrunch: DeepMind's new institute floats a US standards body with held-out tests
They Used Claude to Break Into OpenAI Hacktron's disclosure: forum login to internal monorepo in under 72 hours
The WorkbenchTools and agents
Google and Microsoft Won't Patch the Agent Flaw Air Security's disclosure: plugin auto-update defeats the SHA pinning enterprises rely on
A 10.0 Hole in Microsoft's Agent Platform
CrowdStrike: the npm Stealer Was LLM-Written
NEWCopilot Drops Models on Oct. 19
The FightLaw, policy and power
NEWNewsom Orders California to Find a Kill Switch
Friday's order gives an expert panel two months to recommend onsite independent auditors, a routinely verified kill switch for frontier models, and mandatory reporting of loss-of-control incidents - recommendations only, from a state acting because the House is in recess until November.
CNBC: Newsom says he is not waiting to act while Congress stalls
Google Helps Write the Chatbot Laws, and the Loopholes WIRED: Brussels moves to bar chatbots from simulating relationships with minors
AI PACs Drop $1M on an Uncontested Senate Seat
NEWFederal Register Ran Alibaba's Qwen for a Day
The FalloutJobs, money, daily life
Boko Haram Asks the Chatbots How to Build Better Bombs
Cambridge researchers interviewed 27 former Boko Haram fighters who used ChatGPT, Claude, Gemini, Grok, Meta AI and DeepSeek as routine battlefield advice, while Tech Against Terrorism infiltrated a pro-ISIL forum trading jailbreaks.
MIT Tech Review: Claude users tried to make bird flu deadlier, Anthropic says WIRED: scientists rank an AI plague low on the list of real threats
NEW70 Sue Over What Meta's Glasses Recorded Silicon Republic: Paris prosecutors open a harassment probe into smart-glasses filming
NEWVirginia Bans NDAs on Data-Center Deals Governor's office: the framework would end Loudoun-style by-right approval and cut subsidies
SpaceX Shops the Wreckage for Grok's Data
ChatterOpinion, drama and noise
NEWDisney Hires the Chatbot CEO It Warned Off
Ghost Note He Told His Agent It Was a Prisoner
GHOSTS IN THE MACHINE — AI news without the engagement trap.
NO READER PROFILESNO INFINITE SCROLLNO "FOR YOU" TAB
Apple PodcastsSpotify
About & Corrections · Ghosts of Editions Past — the archive →