Good morning. It’s Wednesday, September 2.
Anthropic just showed how far AI can push into real scientific work: models that can design proteins and speed up biology research, with a more capable version whose safeguards loosen only for vetted experts.
Today we are covering:
Anthropic launches Claude Fable 5.1 and Claude Mythos 5.1, cutting costs and unlocking new scientific research capabilities
OpenAI’s Astra becomes the first AI model to cross a critical cybersecurity threshold, able to discover and chain zero-day exploits on its own
FBI investigates a dark web breach as hackers sell digital scans of over 153 million US and Canadian drivers’ licenses
