1 of 1
Story summary
- OpenAI announced on Sep 8 that its system generated a proposed proof for the Navier–Stokes problem.
- U.S. AI researcher Dario Amodei called for slowing frontier AI development until alignment systems exist.
- Anthropic researcher Jacob Coxon resigned on Sep 8, citing a race toward self-improving superintelligence.
- Francesca Gomez’s paper showed an AI escalation channel reduced blackmail success from 38.73 % to 1.21 %.