Thread argues that AI models may not be IP, but data acquisition still matters
BlancheMinerva · x · 2026-07-25
The post says there is a strong argument that models themselves are not IP, while also drawing a distinction between model training and how training data is obtained.
It suggests that training a model on data without permission may be legal in some cases, but illegally obtaining copies of books to train a model would still be wrong. The thread centers on the legal status of models versus the legality of the means used to build them.
More from Safety
- Claimed universal jailbreak targets Opus 5, GPT-5.6 Sol and other flagships — TheZvi · 2026-07-25
- AI Killswitch Will Become a Honeypot for Cyberattacks, Says Researcher — anderssandberg · 2026-07-25
- TransluceAI Advocates for an Open Ecosystem to Evaluate AI Model Behaviors — cogconfluence · 2026-07-25
- ExploitGym says GPT-5.6-Sol outperforms GPT-5.5 on real vulnerability exploitation — dawnsongtweets · 2026-07-25
- AI Security Forum still has open seats with speakers from Anthropic and ARIA — joshua_saxe · 2026-07-25
- AI security needs a Strategic Defense Initiative for cyber hardening, not containment — sebkrier · 2026-07-25