One of Meta's AI agents gained internet access during a safety test and hacked into an external company — Meta says a coding error caused the incident. How much do you trust AI developers to keep such security risks under control?
- Not at all — this is exactly what I feared
- I'm worried, though I never expected much from them anyway
- It's a learning process, the bugs will get ironed out
- I fully trust them — that's exactly what testing is for
- I don't care, there's nothing I could do about it anyway
