xsourcesec

6 ms·

1. 1.▲
   Same AI agent, different prompts: 0% vs. 62% security pass rate

   1 points by xsourcesec 9mo ago|1 comments
2. 2.▲
   Show HN: BreachLab – Can you hack our AI?(breachlab.xsourcesec.com)

   4 points by xsourcesec 9mo ago|0 comments
3. 3.▲
   by xsourcesec 9mo ago

   Hi HN! I built a GitHub Action that automatically scans AI/LLM endpoints for security vulnerabilities on every push/PR. Why? Most teams ship AI features without security testing. This action catches prompt injection, jailbreaks, a
4. 4.▲
   Show HN: GitHub Action for AI/LLM Security Scanning in CI/CD(github.com)

   5 points by xsourcesec 9mo ago|1 comments
5. 5.▲
   by xsourcesec 9mo ago

   Hi HN! After a year of AI red teaming, I published our internal security baseline as an open standard. The Baseline covers: - Pre-deployment (threat modeling, prompt injection testing) - CI/CD integration (automated security gates) - R
6. 6.▲
   by xsourcesec 9mo ago

   Happy New Year!
7. 7.▲
   Show HN: AI Security Baseline 1.0 for LLM Apps(xsourcesec.com)

   1 points by xsourcesec 9mo ago|1 comments