Probing Stylistic Appropriation using Large Language Models: An Evaluation Framework for Copyright Infringement under EU Law
Researchers have identified a critical gap between how LLMs are currently tested for copyright compliance and what EU law actually protects. While existing safeguards focus on detecting verbatim memorization, EU doctrine covers broader infringement including stylistic mimicry, narrative structure, and character development. The team introduces PSALM, a framework that uses LLMs themselves as evaluators across computational, stylistic, and content dimensions while accounting for legal exceptions like parody and fair quotation. This work signals growing pressure on AI developers to align technical compliance mechanisms with actual legal liability, particularly in jurisdictions with expansive copyright standards.
Modelwire context
ExplainerThe deeper issue PSALM surfaces is jurisdictional asymmetry: EU copyright law protects the expression of ideas at a level of abstraction that most AI safety tooling was never designed to reach, meaning a model can pass every existing compliance check and still expose its developer to infringement claims in European courts.
This connects most directly to the Anthropic model reinstatement coverage from July 1 (the TechCrunch and Verge pieces on Mythos and Fable). Those stories framed regulatory friction as primarily a U.S. export-control problem, but PSALM points to a separate and less negotiable constraint: EU intellectual property doctrine, which no administration policy reversal can waive. While the Anthropic stories suggest that policy-driven model gating may be softening in the U.S., the copyright exposure PSALM identifies runs in the opposite direction, toward tighter legal accountability in a major deployment market. The two pressures are largely independent, which means labs navigating one successfully may still be walking into the other.
Watch whether any EU member state regulator or rights-holder organization cites a framework like PSALM in formal enforcement guidance within the next 12 months. That would signal the research has crossed from academic proposal into actual legal infrastructure.
Coverage we drew on
- Trump drops restrictions on Anthropic’s Mythos and Fable models · TechCrunch - AI
This analysis is generated by Modelwire’s editorial layer from our archive and the summary above. It is not a substitute for the original reporting. How we write it.
MentionsPSALM · Large Language Models · EU Copyright Law
Modelwire Editorial
This synthesis and analysis was prepared by the Modelwire editorial team. We use advanced language models to read, ground, and connect the day’s most significant AI developments, providing original strategic context that helps practitioners and leaders stay ahead of the frontier.
Modelwire summarizes, we don’t republish. The full content lives on arxiv.org. If you’re a publisher and want a different summarization policy for your work, see our takedown page.