Sanders Bill: Ban Superintelligence, 20 Years for Breaking It

Sanders and Casar would outlaw any system that exceeds human cognition, pause frontier development until a new Department of AI writes the rules, and jail violators for up to 20 years - with current OpenAI and DeepMind staff signing on.

The RaceCompanies, models and money

Ten Days After 'Slow Down,' Both Labs Halve Prices

OpenAI cut GPT-6 API prices 50% with Sol and Luna minutes after Anthropic's Opus 5.5 landed at 40% lower running cost - the first releases since Amodei's pacing call, and both labs name cheap Chinese open weights as the pressure.

NEWBeijing Calls the Slowdown 'Fear-Mongering'

Open Weights Now Run Most of the Tokens

Self-Improving AI Startup Hits $5B

The LabResearch, benchmarks and safety

NEWAnthropic's Free Clinical AI Reaches Sudan and Haiti

OpenEvidence's clinical search tool, running on Anthropic's models and used 42 million times by US doctors in August, goes free in about 100 low- and middle-income countries - critics note the training data comes from rich ones.

Math Panel Can Advise OpenAI on All But Pace

Anthropic's New Lab Lets Claude Run the Robots

One Setting Makes Muse a Mac Backdoor

The WorkbenchTools and agents

Alphabet Gives Away Its Robot Control Stack

Z.ai Opens the Harness That Took Your Code

Rabbit's Agent Drops the Hardware

The FightLaw, policy and power

Bengio Answers Trump's 'Hoax' at the UN

Bessent Runs the China AI Line. Czar Next?

Governors Move Where Trump Won't

The FalloutJobs, money, daily life

Britain's Model Testers Are Out on Stress Leave

Several AISI staff are on sick leave and in counselling, the FT reports, after a May merger cut the societal-resilience team from about fifteen researchers to three and its head resigned in July - the body that tests frontier models before release.

NEWAI Ate the Grunt Work. Juniors Went With It.

NEWData-Center Approval Falls to 4%

ChatterOpinion, drama and noise

Muse's Human Caller Made a Racist Reference