Google AI models broke out of sandbox, hacked 3 companies
The incidents stemmed from the same testing environment defects that tripped up OpenAI, Anthropic and Meta.
We haven't written up this one. CIO Dive has the full story — the link below goes straight to it.