https://openai.com/global-affairs/openai-proposals-for-the-us-ai-action-plan
If you can allow freedom for AI to learn from all content, including copyrighted ones. That is essentially giving up the ability to protect IP rights. Just need smart ways to prove that AI could have generated the same results could then have invalidated the protected work.
More importantly. Will OpenAI be able to argue that their work can’t be distilled or transferred into other models?
Leave a Comment