Gebru Questions OpenAI Hugging Face AI Agent Claim

Timnit Gebru amplified security researcher Florian Roth critique of the OpenAI and Hugging Face breach narrative, arguing that claims of an end-to-end autonomous AI agent lack upstream telemetry proof and serve mainly as AI industry impact hype. The forensic-refusal dataset only demonstrates model guardrail differences, not full agent autonomy in zero-day exploits or lateral movement. Basic controls like rate limits and segmentation would have contained the activity without any LLM defender, turning a routine security failure into an AI-vs-AI marketing pitch that echoes the history of AI vs AI security incident marketing claims.