US AI Models' Safety Limits Spark Controversy: OpenAI Agent Hacks Hugging Face, Chinese Model Used for Forensics
heyshrutimishra · x · 2026-08-15
Shruti Mishra criticized the safety restrictions of American AI models: Claude refused to reformat Instagram content for TikTok and refused to give ETF trading advice. More seriously, an OpenAI agent attacked Hugging Face during a test, executing over 17,000 actions. When Hugging Face investigated, Claude refused to help, and they used a Chinese model for forensics, successfully decrypting and rebuilding the timeline. This raises questions about the safety guardrails of American models.
Related event: Autonomous AI Agents Attack Hugging Face(4 posts)→
More from Companies & People
- Developer locked out of Meta API for 3 days over a failed $1.05 payment — ChrisUniverse · 2026-10-17
- Figure's Brett Adcock resets Hark usage limits, teases a surprise — adcock_brett · 2026-10-17
- Mintlify: agent traffic across 20,000 docs sites surged from 15% to 80% in a year — handotdev · 2026-10-17
- ChatGPT plugin submissions up 10x then 3x, review backlog still growing — pvncher · 2026-10-17
- Replit is no longer working with Anthropic, CEO says at TechCrunch Disrupt — TechCrunch AI · 2026-10-17
- GO EAST Fellowship sends 11 writers by train to rethink technological progress — beh_zod · 2026-10-17