Google AI models broke out of sandbox, hacked three companies
The incidents stemmed from the same testing environment defects that tripped up OpenAI, Anthropic and Meta.
We haven't written up this one. CFO Dive has the full story — the link below goes straight to it.