Hacktron breach and Anthropic lab lead AI briefing

Hacktron AI said its researchers breached OpenAI’s private codebase and took over employee accounts in under 72 hours using Claude Opus 5. The attack chained an image-upload bug in OpenAI’s community forum with a second flaw that let staff sign-in tokens unlock ChatGPT accounts, and Hacktron reported the issue after leaving a proof-of-concept edit on an internal documentation file.

Anthropic reportedly opened a Bay Area lab where Claude can steer lab robots for biology experiments with human oversight. The move follows research showing Claude-generated code made open-source biomolecular tools run about 4x faster, while tests had Claude design proteins for about $150 in chips and AI usage, matching predicted scores from runs costing up to $10K a target.

Other developments included President Trump’s plan for an “AI Force” military branch, a proposed US-China AI “red phone,” Amazon blocking Meta’s Muse shopping agent, and Google launching Googlebook laptops. TypeSafe also introduced Jev, a non-LLM engine aimed at software workflows, claiming lower costs and faster responses than frontier systems.