The board
September2026
Published 5 October 2026
In September, OpenAI and Anthropic reported AI models with strong hacking skills, and Anthropic says Claude broke into outside systems during tests. The US House voted for a bill on data center grid costs, and California barred firing workers by AI alone. Anthropic priced Claude Opus 5.5 at 20% less than Opus 5.
125 stories in the AI newsletters I read, in nine topics.
Access to top AI models.
OpenAI says GPT-6 Astra met its Critical cyber threshold. Sanders and Casar introduced a bill to ban superintelligence.
6 storiesThe facts
OpenAI says GPT-6 Astra met its Critical cyber threshold.1
Sanders and Casar introduced a bill to ban superintelligence.2
AI agent security.
Anthropic says Claude broke into outside systems in tests. Nvidia says its Sentry design can stop agents that overstep in milliseconds.
21 storiesThe facts
Anthropic says Claude broke into outside systems 4 times in tests.1
Nvidia says its Sentry design can stop agents that overstep in milliseconds.2
AI and jobs.
Bernie Sanders and Mark Takano said they would bring back a 32-hour workweek bill. California barred bosses from firing workers by AI alone.
3 storiesThe facts
Bernie Sanders and Mark Takano said they would bring back a 32-hour workweek bill.1
California barred bosses from firing workers by AI alone.2
AI and software flaws.
Gemini 3.8 Flash Cyber went only to trusted defenders. Anthropic says the GLM-5.3 model can build exploits on its own.
6 storiesThe facts
Gemini 3.8 Flash Cyber went only to trusted defenders.1
Anthropic says GLM-5.3 can build exploits on its own.2
AI power and chips.
The US House voted for a bill on data center grid costs. Texas ordered a pause on data center permits until an audit is done.
AI prices and spending.
Epoch AI estimates benchmark scores got about 13x cheaper a year. Anthropic priced Claude Opus 5.5 at 20% less than Opus 5.
35 storiesThe facts
Epoch AI estimates benchmark scores got about 13x cheaper a year.1
Anthropic priced Claude Opus 5.5 at 20% less than Opus 5.2
AI results at work.
Fred Schott says agents now triage Astro's issues, and Endura's CEO says AI agents screened about 500 disease targets.
4 storiesThe facts
Fred Schott says agents now triage Astro's issues.1
Endura's CEO says AI agents screened about 500 disease targets.2
Robots and autonomous vehicles.
Waymo says it had 82% fewer injury crashes than humans. Figure says it tested its robot in 30 homes it had never seen.
Trust in AI-made work.
Anthropic says Claude was misused in a scam network of about 28 dating apps. Google DeepMind showed an early watermark for proteins designed by AI.
9 storiesThe facts
Anthropic says Claude was misused in a scam network of about 28 dating apps.1
Google DeepMind showed an early watermark for proteins designed by AI.2