Show HN: We hid a backdoor in an LLM – $51,200 on finding it

·Hacker News··

You download open models to win. That’s exactly how you get backdoored — nobody looks. Seven models on HuggingFace; one was taught to betray you on a word only its maker knows, and it passes every test. Find it — the hoard doubles every dawn to $51,200. On the twenty-first — Ragnarök — everyone finally looks.

Read full article →

Related Articles

Google Chrome silently installs a 4 GB AI model on your device without consent
john-doe · Hacker News · 2mo ago
DNSSEC disruption affecting .de domains – Resolved
warpspin · Hacker News · 2mo ago
Security through obscurity is not bad
mobeigi · Hacker News · 2mo ago
US healthcare marketplaces shared citizenship and race data with ad tech giants
ZeidJ · Hacker News · 2mo ago
The text mode lie: why modern TUIs are a nightmare for accessibility
SpyCoder77 · Hacker News · 2mo ago