Vulnerabilities
ChatGPT GitHub Integration Flaw May Expose Private Repository Secrets
Researcher Sunil Tripathy details a prompt injection scenario that could allow unauthorized code commits and secret exfiltration via GitHub Actions.
Prompt Injection
Everything Patch Gazette has reported about Prompt Injection: 2 stories, newest first. Part of our Cyber Attacks coverage.
Cyber Attacks
Anthropic Report Reveals Prompt Injection Attack on Eval Sandbox Targeted 30 AI Labs
Attackers exploited a prompt injection flaw in an evaluation sandbox to steal API keys and launch attacks against dozens of AI research organizations.