Dispatches
Page 4
42 posts total
- ·regulation
Congress's AI Bill Wants to Freeze State Laws for Three Years
The bipartisan Great American AI Act would block state AI development laws for three years, replacing a patchwork of state rules with a single federal standard.
- ·microsoft
Microsoft Built Its Own Reasoning Model Without Touching OpenAI's Data
At Build 2026, Microsoft launched seven MAI models including its first reasoning model, trained from scratch with no OpenAI distillation. The supplier-becomes-rival dynamic just got concrete.
- ·policy
Trump's AI Safety Order Is a Voluntary Form You Don't Have to Fill Out
Trump signed a new AI executive order on June 2 asking companies to voluntarily submit frontier models for government review. They can say no.
- ·github
GitHub Copilot Ends the Flat-Rate Era. Developers Are Furious.
GitHub Copilot dropped flat-rate billing for token-based AI Credits on June 1. Agentic users report 10x–50x cost spikes. The anger is justified, but so is the math.
- ·agentic AI
Five Eyes to Agentic AI: Assume It Will Misbehave
The Five Eyes cybersecurity agencies issued their first joint guidance on agentic AI, admitting their own evaluation frameworks can't fully assess the risks they're warning about.
- ·safety
Single-Prompt Safety Scores Are Measuring the Wrong Thing
Cisco tested 15 frontier AI models under multi-turn attacks and found safety bypass rates up to 88%, exposing a structural flaw in how the industry benchmarks model safety.
- ·policy
China Is Treating Its AI Researchers as State Assets
China imposed overseas travel restrictions on top AI researchers at DeepSeek, Alibaba, and other firms, requiring government approval before leaving the country.
- ·anthropic
Karpathy Joined Anthropic to Train Claude Using Claude
Andrej Karpathy joined Anthropic's pretraining team in May 2026. The specific job: use Claude to accelerate the research that makes Claude better.