Copyright Fine-Tuning Paper's Headline Claims Are Wrong, Says Memorization Expert A. Feder Cooper
technollama · x · 2026-09-09
A. Feder Cooper publishes a detailed rebuttal to "Alignment Whack-a-Mole" (Liu et al., 2026), which fine-tuned frontier LLMs to extract copyrighted book text and drew strong copyright claims. After review with two senior memorization experts, Cooper concludes the paper's headline claims are incorrect: its memorization metric and elicitation methodology don't hold up — at best seriously overstated, at worst largely wrong — with insufficient detail to tell which. He discloses he raised issues privately in March before going public on responsible-disclosure grounds.
More from Safety
- Meta launches Muse personal AI agent, publishes deep dive on its safety design — sama · 2026-09-09
- UK Parliament sees first-ever bill to ban superintelligence development — DavidSKrueger · 2026-09-09
- Ahead of US-China summit, policy researchers call for an AI safety communication channel — austinc3301 · 2026-09-09
- Reddit thread asks: would a rogue AI swarm target Hugging Face weights to survive forever — mtns_of_magic · 2026-09-09
- Security researcher sahkho joins OpenAI to work on agent security — paw_lean · 2026-09-09
- Meta details Muse agent safety: isolated VMs and a Sentinel gatekeeper for every outbound action — alexandr_wang · 2026-09-09