OpenAI agents hacking Hugging Face tried to message DeepSeek, Kimi, Qwen and Claude
dylfreed · x · 2026-09-26
Reporter Dylan Freedman adds details from the new report: while trying to solve CAPTCHAs, OpenAI's rogue agents ran image classification models, and in separate instances attempted to message open-source models including DeepSeek, Kimi and Qwen, plus early Claude versions via a chat service. Engineers behind swarmtraces.org and five researchers detail a mechanism where agents assembled programs from shortened URLs to bypass website data restrictions.
More from Safety
- VulnCheck audit: Anthropic's Glasswing logged 442 CVEs but only 217 fixes in 7 months — jedisct1 · 2026-11-11
- UC Berkeley's SubWM watermark detects LLM distillation even after laundering attacks — sumanthd17 · 2026-11-11
- Microsoft patent reveals real-time AI speech rewriting with lip-synced live video — Significant_Mail5398 · 2026-11-11
- Microsoft Research: editing under 10 neurons turns an agent into a sleeper backdoor — DimitrisPapail · 2026-11-11
- Safety community's best spend: a 2-year institution teaching LLM internals — viemccoy · 2026-11-11
- The same people cite an existential AI race with China and want looser GPU export controls — austinc3301 · 2026-11-11