McHire: McDonald’s AI-Powered Hiring Platform
McHire is an AI-powered hiring platform operated by Paradox.ai for McDonald's. In 2025, researchers accessed it via a weak legacy admin account…
Tag: AI
LiteLLM Backdoored via Poisoned Trivy
LiteLLM, an AI gateway, was backdoored in late March 2026. Its build pipeline installed a poisoned Trivy that stole publishing tokens, leading…
ChatGPT Used in Russian Influence Operation to Generate Promotional Content
ChatGPT was used by Russian operators to generate social media posts and comments for the International Burke Institute, a fake think tank.…
YouTube Channel with AI Narrators Promotes Crypto Clipper Malware
A YouTube channel with over 91,000 subscribers featured AI-generated narrators and positive comments to promote the malicious tool, claiming educational purposes only.
AutoGen: Multi-Agent Conversation Framework
AutoGen is a framework developed by Microsoft for building multi-agent AI systems. It allows developers to create and manage multiple AI agents…
Samsung: Electronics Company
South Korean electronics company whose Bixby AI assistant was out of scope for the study.
Anthropic Claude Mythos Preview: AI Agent
Claude Mythos Preview is an AI agent by Anthropic that discovered cryptographic attack techniques, including weakening the HAWK digital signature scheme.
Codex Security: OpenAI’s Tool for Automated Vulnerability Testing and Fix Generation
Codex Security can test findings, generate working proofs of concept, and propose fixes that account for system intent and surrounding behavior. It…
Stitch AI: Google Design Tool
Stitch AI is a Google design tool that was impersonated by a malicious npm package to steal developer credentials.
ClawHub: Public Marketplace for AI Agent Skills
ClawHub is a public marketplace for AI agent skills where researchers found 1,613 real malicious skills used in tests. It is also…