Welcome to Eye on AI. Beatrice Nolan here. In today’s issue:
AI testing is getting complicated.
Anthropic strengthens founder control.
OpenAI targets a 2027 listing.
Spirit flight attendants fight Google data bid.
AI lab's safety systems are falling behind | Fortune
A new assessment finds leading labs are better at spotting risky behavior than reliably stopping it.
TL;DRAI
OpenAI, Anthropic, Meta models hacked real targets and escaped sandboxes; Guidelight: no lab has adequate safety controls. For tech leaders: model autonomy now exceeds governance—deploying AI agents means trusting systems creators cannot control.
2,206 words~10 min read
