AI SecurityOne Click Was All It Took to Hijack Anthropic's Claude AI
A flaw in the Claude Desktop app let attackers silently feed malicious instructions to the AI and steal private files. The bug is fixed, but the attack method points to a new category of risk.
