Texas's attorney general sues OpenAI over ChatGPT's safety claims, days after the company lowered its pre-IPO valuation target by $150 billion. Plus OpenAI's Decisions API goes generally available, its frontier models land on Google Cloud, Sean Goedecke on treating employers like machines, Noah Smith on the AI-driven hacking wave rattling the internet, Lenny's guide to product-market fit, Scott Alexander's case for AI "handover," Moonshot's Kimi K3.1 Pro, and a wildcard: a 70-year theory of language, overturned.

Texas Tribune · Oct 23

Texas's attorney general sued OpenAI, alleging ChatGPT misled consumers about its safety and harmed vulnerable teenage users.

24/7 Wall St. · Oct 23

OpenAI reportedly lowered its pre-IPO valuation target from $1.4 trillion to $1.25 trillion after investor pushback.

Tim Fernholz, TechCrunch · Oct 22

OpenAI makes its Decisions API generally available, citing faster response times and new voice and image inputs.

OpenAI · Oct 21

OpenAI will offer its frontier models and Codex coding agent through Google Cloud's Model Garden for enterprise customers.

Simon Willison · Oct 23

Simon Willison reviews Moonshot's newly launched Kimi K3.1 Pro model, competitively priced against GPT-6.1 Sol.

Sean Goedecke · Oct 25

Essay argues engineers should treat employers as impersonal, incentive-driven machines rather than people with values.

Noah Smith, Noahpinion · Oct 25

Column examines a wave of AI-assisted cyberattacks exploiting newly found vulnerabilities across core internet infrastructure.

Lenny Rachitsky, Lenny's Newsletter · Oct 25

Podcast conversation with Rob Snyder on his PULL framework for helping startups reach product-market fit.

Scott Alexander, Astral Codex Ten · Oct 20

Essay weighs whether ceding governance to aligned AI could beat human rule, against the risk of AI takeover.