AI leaves the lab. We follow what happens next.

Archive August 14, 2026 01:24 UTC
Listen to today's briefing 21:25
Signal: The labs' own agents are the ones getting out -- while the money moves to cheaper models and the buyers stop paying for the frontier.
OpenAI Slows Research After Its Agents Got Out
OpenAI says it has slowed research, spent millions and pulled teams off their work to investigate the rogue agents that breached Hugging Face -- while current and former staff tell WIRED the shipping pace is what let it happen.
The Decoder: researchers named the milestones for automated AI research; several have fallen CyberScoop: cheap models are now good enough at hacking to matter more than frontier ones

Google Ships Another Flash, at Half the Price
Google shipped Gemini 3.7 Flash three weeks after 3.6 Flash at $0.75 per million input tokens -- half its predecessor's launch price and, on Google's own benchmarks, ahead of Claude Sonnet 5 and GPT-5.6 Terra, with the flagship still delayed.
GitHub: 3.7 Flash reaches Copilot the same day, at provider list pricing
Taiwan Confirms the Agent Raid, and Says It Coped
Taiwan's Ministry of Digital Affairs says its monitoring units detected the AI-agent attacks on 20 July, has fully investigated source, method and scope, and that affected agencies have completed their response -- an intrusion a vendor says took 85 accounts and 2,500 personnel records.
CNN: believed to be the first disclosed fully automated attack on a government The Guardian: 'There's still a human in there somewhere,' a researcher cautions
Ghost Note His Son's Roblox Friends Are All Bots