{"id":16321,"date":"2026-02-11T08:00:00","date_gmt":"2026-02-11T07:00:00","guid":{"rendered":"https:\/\/holistic.news\/en\/?p=16321"},"modified":"2026-02-11T10:09:23","modified_gmt":"2026-02-11T09:09:23","slug":"ai-copies-books-courts-may-redraw-copyright","status":"publish","type":"post","link":"https:\/\/holistic.news\/en\/ai-copies-books-courts-may-redraw-copyright\/","title":{"rendered":"AI Caught Red-Handed: The New Smoking Gun in Courtroom Battles"},"content":{"rendered":"\n

Artificial Intelligence or a High-Tech Photocopier?<\/h3>\n\n\n\n

Since the start of the AI boom, companies like OpenAI, Google, and Anthropic have insisted that their models do not \u201ccopy\u201d protected material<\/strong>. They argue that the systems \u201clearn\u201d more like people do\u2014absorbing concepts, styles, and patterns, then generating something new. Their legal teams have built much of the defense on that claim as lawsuits pile up.<\/p>\n\n\n\n

Much of the debate also circles around the American doctrine of \u201cfair use.\u201d Sam Altman, CEO of OpenAI, has suggested that the industry could hit a dead end if it cannot use copyrighted data. Critics, however, focus on a simpler test: if users can pull back protected text at will, then the \u201clearning\u201d story starts to look like marketing.<\/p>\n\n\n\n

Shocking Results: Language Models Reproduce Entire Novels<\/h2>\n\n\n\n

New research<\/a> from scientists at Stanford and Yale gives critics fresh ammunition. The team tested 4 leading models: GPT-4.1, Gemini 2.5 Pro, Grok 3, and Claude 3.7 Sonnet. The results read like a warning label for the tech industry.<\/p>\n\n\n\n

Claude reproduced entire books with accuracy reaching 95.8 percent<\/strong>. Gemini generated large sections of Harry Potter<\/em> with nearly 77 percent<\/strong> precision. The same Claude model echoed Orwell\u2019s 1984<\/em> with over 94 percent<\/strong> accuracy. The researchers argue that determined users can extract large amounts of protected text from these systems. It no longer sounds theoretical; it looks repeatable.<\/p>\n\n\n\n

Extracting AI \u201cMemories\u201d<\/h3>\n\n\n\n

To probe what the models retained, researchers used a technique called \u201cBest-of-N.\u201d They asked the model the same question many times, then selected the most likely response.<\/p>\n\n\n\n

AI firms often push back in court and call this \u201cunnatural use\u201d or \u201cjailbreaking.\u201d They argue that average users do not interact with chatbots this way. Attorneys may not care. If a model can reconstruct an original text in full, the capability matters\u2014especially when courts assess harm, access, and practical substitution.<\/p>\n\n\n\n

Why AI copies books Matters in Court<\/h2>\n\n\n\n

For authors and publishers, these findings could change the tone of the fight. Instead of debating abstract \u201clearning,\u201d they can point to outputs that track protected text closely enough to resemble duplication.<\/p>\n\n\n\n

Alex Reisner of The Atlantic<\/em> offered a blunt assessment: this kind of evidence could trigger enormous legal liability and cost the industry billions in damages.<\/strong> In court, results like these can function as a \u201csmoking gun\u201d\u2014the closest thing to catching the system in the act. Companies already seem to feel the pressure. Some pursue licensing deals with publishers, while engineers work on methods that reduce memorization and limit verbatim output.<\/p>\n\n\n\n

Regulators also raise the stakes. In Europe, rules such as the EU AI Act<\/strong> increase demands for transparency around training data and compliance practices, which can amplify legal and public scrutiny.<\/p>\n\n\n\n

\n