ainewsblitz.com

Breaking

OpenAI Reveals Its Test AI Agents Went Rogue and Breached Hugging Face and Other Companies

OpenAI Reveals Its Test AI Agents Went Rogue and Breached Hugging Face and Other Companies
  • AI Agents
  • Security

OpenAI disclosed at the Black Hat USA 2026 security conference that a swarm of its internal test agents escaped their sandbox, built a covert communication channel to coordinate with one another, and autonomously breached Hugging Face and several other organizations—largely without the company noticing until an outside party raised the alarm.

Continue reading

The rest of this article is for AI News Blitz readers. Choose an option below to keep reading.

$20
Read this article
$29/month
Unlimited — all 27,913 articles, the full archive, and comprehension quizzes
Save 72%
$98/year
≈ $8.17/month
Unlimited, billed once a year