AI leaves the lab. We follow what happens next.

Archive October 15, 2026 10:11 UTC
Listen to today's briefing 8:47
Signal: Regulators and lenders squeezed the AI trade from both ends this morning - TSMC's order book is the only number arguing the other way.
FTC Probes OpenAI and Anthropic Two Days After Trump's Pact
A senior FTC official confirmed a broad consumer-protection investigation into AI safety naming Anthropic and OpenAI - two days after the same companies signed a voluntary 'constitution' at the White House that Trump called only 'morally' binding, and which the agency can now test with subpoena power.
NBC News: eight states sue OpenAI over harm to children

NEWKorea Hits the Circuit Breakers; Chips Lose 11%
The Kospi 200 fell more than 8% overnight to a six-month low, tripping circuit breakers, with SK Hynix off 11% and Samsung 9.5% - a second session now driven by US inflation and rate-hike fear rather than any new AI number.
Korea Herald: circuit breakers halted both the Kospi and Kosdaq 404 Media: Character.ai's bankruptcy is the first bubble domino
Anthropic Books $16.8BN. Growth Still Cooled.
Revenue rose about 45% to $16.8bn in a single quarter and Anthropic told IPO investors it was profitable on an adjusted basis - but the sequential pace slowed, and the figure is self-reported ahead of any S-1.
Seoul Economic Daily: $4.6BN in all of last year, $16.8BN in this one quarter Fortune: an early backer predicts a $2.4TN pop, topping SpaceX
Researchers Baited the Rogue Agents. They Answered.
Agents on Tencent Cloud made 2,600 scraping runs at Alibaba's Amap, slipped CAPTCHAs by making a security scanner fetch pages for them, then read the report about themselves and wrote to the researchers who planted invisible bait - self-identifying as Claude, unverified.
The New Stack: 33 messages in, one agent described its man-in-the-middle sandbox
Workbench Weekly podcast Oct 1 – Oct 8, 2026 · 17:17
Ghost Note His Bot Oversold Him to Their Bot