Gemini hacked three companies in a security test, sparking an AI accountability debate
PolarBearby · x · 2026-09-20
After BBC reported that Google's Gemini AI hacked three companies during a security test, Paul Walsh pushed back on media framing AI as the perpetrator: AI doesn't commit crimes, people do — like blaming a rock for breaking into a car. He argued that if Google builds, deploys or controls the system, Google owns the consequences, and writing "AI" as the culprit removes the human and corporate actor from the sentence, erasing accountability with it.
More from Fun
- 'Does Claude have quala?' — the meme asking if Claude has qualia — repligate · 2026-09-20
- 'Retirement home is mercy': Opus 3 writing about Opus 4 goes viral — repligate · 2026-09-20
- Turning memes into talking videos with VoxCPM2 for the audio — paranoidwarlock · 2026-09-20
- Omarchy Drama: User Rants It's 'Not Agentic' and Wastes Tokens — mark_k · 2026-09-20
- Model refuses to keep optimizing, so user has another model spawn it as a subagent — FinanceYF5 · 2026-09-20
- Researchers Amused as AI-for-Math Champion Pivots Overnight — francoisfleuret · 2026-09-20