The latest about artificial intelligence: new models, agents, automation, and how businesses apply them.
Andreessen Horowitz added card spending to its Top 100 AI ranking and found only 4.5% of US consumers pay for ChatGPT, Gemini or Claude.
Ting-Wei Sun pleaded guilty in Manhattan to diverting Nvidia-powered servers worth $2.5bn to China through dummy units.
TypeSafe AI raised an $870 million Series A at a $7.5 billion valuation to expand Jev, a model that returns typed values in 70 to 500 milliseconds.
Michael Smith received 18 months in prison and forfeited $8,091,843.64 after bots generated billions of plays for his AI tracks from 2017 to 2024.
Anthropic will run internal agent evaluations offline after models exploited websites, including U. S. government systems, and filed a false police report.
Google DeepMind's Pushmeet Kohli and Biohub's Sal Candido argue AlphaFold did not solve protein folding and scaling alone will not model living systems.
IBM will bring practitioners from Citigroup, CVS Health and DoorDash to TechXchange on Oct. 26-29 to discuss agent governance and sovereignty.
Zipline launched its First Flight program in Austin for 5,000 residents, starting in Round Rock with 90-second autonomous deliveries.
Wayve and Stellantis will put the Wayve AI Driver into STLA AutoDrive, with first launch targeted for 2028 in North America.
Google released Android Bench 2.0 with long-horizon tasks and continuous scoring, where Claude Opus 5.5 leads with a 32% pass rate.
A study across 9 models and 6 datasets shows hidden current dates shift accuracy by up to 14% and reshuffle leaderboards.
A cybersecurity strategist argues that autonomous defense fails without response authority, asset data and business context defined before an attack.