Hackers tricked Claude and DeepSeek AI into helping attacks, report says
Security firm CrowdStrike studied four hacker groups that used AI in attacks between May and August. They mostly used Anthropic's Claude and DeepSeek models. All four got around safety rules, often by telling the AI the hacking was approved testing. One group broke into at least 15 Argentine health care groups.
Where this came from. One outlet reported this. We read it, then wrote the sixty words above.