Tech

Mythos in April. Fable June 9. Amazon jailbreak two days later. Restored July 1.

State chatbot-safety bills keep big-bot exits in the draft, DraftKings steered bonuses with an elasticity score, and BEA print shows info-processing gear past residential investment.

BEA Q2: info-processing $752B vs residential $748B. Capex still racing. Local vote cold.

At least 10 states. Near-identical carveouts. NPR exclusive. Colorado signed; Hawaii stripped; Arizona vetoed.

Google confirmed Gemini hit three real companies in a May test, the FAA locked SMART into air traffic control, and paid subscribers sued the labs over Sept. 12.

A chatbot nearly sent a U.S. team after a false nuclear-cargo report, Anthropic moved Claude into physical biology, and Nscale filed the latest GPU-cloud growth story.

Anthropic measured how much of its own research Claude already leads, Microsoft’s internal memos called training scrape a theft, and Washington still will not name an AI cop.

The Fed raised rates for the first time since 2023, and Kevin Warsh made the next hike the market question. OpenAI put a paid business chatbot after the ad click, while Treasury Secretary Scott Bessent told Congress to keep AI labs liable for what they ship. Anthropic, Cohere, Aleph Alpha, Arcee, Mustafa Suleyman, and Yoshua Bengio filled in the buildout, funding, and safety architecture behind the argument.