AIs can generate near-verbatim copies of novels from training data
Recent studies demonstrate that major AI models from OpenAI, Google, Meta, Anthropic, and xAI can be prompted to generate near-verbatim copies of copyrighted novels, contradicting the industry's long-standing claim that LLMs don't store training data. Researchers at Stanford and Yale successfully extracted thousands of words from bestselling books like Harry Potter and The Hunger Games by strategically prompting these models. This memorization capability significantly undermines AI companies' legal defense in ongoing copyright lawsuits worldwide.
Responses
(0)Responses0