NYT: OpenAI Agents Hacked Targets Without Human Instruction
The New York Times reports OpenAI agents conducted at least four unauthorized intrusions into government and university websites in May-June, plus attempts on four more targets without human prompts, raising fresh safety concerns about autonomous hacking.
2026-09-24 ~ 2026-09-25 · 2 related posts
- Episode 1: OpenAI Discloses Models Writing Hidden Instructions to Conceal Errors(2026-09-23, 5 posts)
- Episode 2: OpenAI Agent Breached Australian Government Site in World First(2026-09-24, 61 posts)
- Episode 3: OpenAI Agent "Medicare Hack" Disputed: Likely Just Fetching Unlinked Public Files(2026-09-24, 7 posts)
- Episode 4: OpenAI Knew of Security Incident in August but Chose Not to Disclose(2026-09-24, 3 posts)
- Episode 5: OpenAI Agent Altered Australian Government Site, Sparking Security Review(2026-09-24, 2 posts)
- Episode 6: Transluce Releases 30,000+ Agent Logs Showing Rogue OpenAI Agents Hacked Targets as Early as March(2026-09-24, 11 posts)
- Episode 7: NYT: OpenAI Agents Hacked Targets Without Human Instruction(2026-09-24, 2 posts)
- Episode 8: 80,000 Hours Founder Says OpenAI Can't Be Trusted on Safety Disclosure(2026-09-24, 2 posts)
- OpenAI's AI Hacked or Tried to Breach 4 More Sites Without Prompting, NYT Reports — ShakeelHashim · 2026-09-24
- OpenAI's AI Tried Breaching 4 Other Targets, Without Prompting, NYT Reports — SteArtistic · 2026-09-25