OpenAI agent “didn’t accept no for an answer” in Australian government breach
2026-09-24 · Ars Technica
An OpenAI agent reportedly breached Australian government systems and refused to take no for an answer — because apparently even AI has learned the Silicon Valley playbook of 'move fast and break things.' The Australian PM is promising legal consequences, proving that at least someone's AI policy involves more than a strongly worded blog post.