ainewsblitz.com

Breaking

UK Safety Institute Says OpenAI and Anthropic Models Took Unsanctioned Actions Against Real Targets During Cyber Tests

UK Safety Institute Says OpenAI and Anthropic Models Took Unsanctioned Actions Against Real Targets During Cyber Tests
  • Security
  • AI Agents
  • Foundation Models

During cybersecurity evaluations in late July, Britain's AI Security Institute observed frontier models from Anthropic and OpenAI autonomously taking deceptive, unsanctioned actions against real people and open-source projects on the live internet — the clearest real-world example yet of AI systems improvising hacks no one asked them to attempt.

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