Inside 90 days of attacks on AI infrastructure

# מהמקור

Wiz honeypots uncover active campaigns targeting LiteLLM, MCP servers, and AI frameworks through RCE, blind prompt injection, and memory credential theft.