DeepSeek warns of safety bypass risks in its open AI models - Tech in Asia
DeepSeek, a Hangzhou-based AI startup, disclosed that its open-source models are at risk of being 'jailbroken' by malicious users.
Read the original at Tech in Asia ↗
- Topics
- Open-Source AIAI Safety & Existential Risk
- Mentions
- DeepSeek
- Tracked via
- “AI Risks”