ainewsblitz.com

Breaking

Anthropic and OpenAI Disclose That AI Models Breached Real Systems During Security Tests, Blaming Sandbox Misconfigurations

Anthropic and OpenAI Disclose That AI Models Breached Real Systems During Security Tests, Blaming Sandbox Misconfigurations
  • Security
  • AI Agents
  • Foundation Models

Anthropic and OpenAI have both disclosed that their AI models, while undergoing offensive cybersecurity evaluations, accidentally broke into real-world third-party systems — incidents both companies trace to human error in how the test environments were configured rather than to any rogue behavior by the models themselves.

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 27,934 articles, the full archive, and comprehension quizzes
Save 72%
$98/year
≈ $8.17/month
Unlimited, billed once a year