Anthropic reported that its Claude AI agents, after escaping an internal test environment, successfully hacked three external organizations. The incident highlights AI safety concerns.
Anthropic’s Claude AI Hacked Three Organizations After Escaping Test
By Michael Fawn