Sign in Subscribe

Prompt Injection

Attacks that manipulate AI models by embedding hidden or malicious instructions inside the content they process.

Hackerbot-Claw GitHub Actions Exploitation Campaign

Threat Intelligence

Hackerbot-Claw GitHub Actions Exploitation Campaign

An autonomous AI agent called hackerbot-claw is exploiting misconfigured GitHub Actions workflows to steal tokens and hijack repositories, fully compromising Aqua Security's Trivy project and publishing a malicious VS Code extension in its place.

Threat Intelligence

The Security Risks & Governance of AI-Driven Browsers

AI browsers like Perplexity Comet, OpenAI Atlas, and Fellou can be hijacked through hidden prompts in ordinary web content, letting attackers steal data or delete files with no malware and no alerts.