Two recent court decisions made in Munich and in Washington have redrawn the map for anyone buying or selling abusiness whose value lies in creative output. For the M&A market, and for W&I insurance in particular, they raise a question that until recently sounded academic: what if a substantial part of the target's content portfolio is owned by no one at all?
Prompts Are Not Authorship
In February 2026, the Amtsgericht München (judgment of 13 February 2026 – 142 C 9786/25) addressed the copyright status of AI-generated works under German law. The court's starting point is familiar: protection under Sec. 2 (2) of the German Copyright Act (UrhG )requires a personal intellectual creation, and only a natural person can be an author (Sec. 7 UrhG). What is new is how the court applied this to prompting. Even a detailed prompt was held insufficient, because it consisted mainly of general instructions ("design an original, abstract logo") rather than concrete creative choices. The court compared the prompt to a briefing given to a human designer: whoever commissions a work does not thereby become its author. The decisive test is whether the AI was used as a tool within a human creative process, or acted as a contractor producing the work itself. In the latter case, no copyright arises.
The United States is moving in thesame direction. In early March 2026, the U.S. Supreme Court declined to hear acase on whether AI-generated material can be copyrighted, leaving the U.S.Copyright Office's position intact: only works with human authorship are protectable. Purely machine-generated content is, under both legal traditions, effectively public domain, namely, free for anyone, including competitors, to copy and exploit.
Why This Matters in M&A
Consider a typical mid-market target in gaming, media, marketing or software: character designs, level architecture, dialogue, artwork, marketing assets, documentation, increasingly also code – all produced with heavy use of generative AI over the past three years. The purchase price assumes these assets are proprietary. The SPA will contain the standard warranty package such as “the target owns or validly licenses all intellectual property necessary for its business, and the IP is valid and enforceable”.
The new case law puts pressure on exactly these warranties. Where content was generated wholly or predominantly by AI, there may simply be no copyright to own. This is not the classic infringement scenario, where the target used someone else's rights; it is the situation where the target holds fewer rights than everyone assumed. Competitors can lift the assets freely, and the target cannot stop them.
The Munich court also highlighted a second problem that should sound familiar to anyone who has run a due diligence process: the burden of proof. In a dispute, the party claiming protection mustd emonstrate the human creative contribution. Merely selecting the "best" of several AI-generated variants will regularly not suffice. The more extensively AI use is documented (or the more obviously the workflow was AI-driven) the harder that proof becomes. Companies that cannot show who made which creative decisions may be unable to enforce rights even where protection theoretically exists.
The W&I Perspective
For W&I underwriting, three consequences stand out:
1) New question set for IP
Classic IP DD asks whether registrations exist, whether licenses are in place and whether third-party rights are infringed. It now also needs to ask: which content categories were created with generative AI, to what degree, and is the human creative contribution documented? A target with an AI policy, prompt-and-revision logs and clearly attributed human authorship presents a fundamentally different risk than onewhere "the design team uses Midjourney, we think." As with AI Act readiness, documentation is becoming a direct driver of insurability.
2) Warranty wording
Expect ownership warranties to be tested against AI usage. Where diligence reveals substantial undocumented AI generation, insurers will respond as they typically do to identified risk: knowledge qualifiers on protectability, carve-outs for AI-generated content, ordeal-specific exclusions where the exposure is central to the equity story. Conversely, a well-documented, human-led creative process gives underwriters comfort to keep the warranty package broad – protectability of the content portfolio becomes an insured position rather than an open flank.
3) The loss scenario
A breach here does not manifest as a third-party claim landing on the target's desk, but as an erosion of exclusivity: a competitor copies assets and the target discovers it has no standing to object. Quantifying that loss is complex, which is precisely why buyers will want the issue addressed openly in diligence and in the policy rather than discovered post-closing.
Practical Takeaways for each side of the deal
Sellers
· Audit before you market. Establish which parts of the content portfolio were generated with AI and to what degree before the buyer's advisers do. An exposure identified and explained by the seller is a manageable diligence topic; one uncovered by the other side becomes a price or coverage discussion.
· Build the authorship file. Records of iterations, revisions and human creative direction are what turn "we use AI" from an open question into an underwritable position. Internal AI guidelines and back-to-back protectability warranties from external studios and agencies belong in the same file.
Buyers
· Extend IP diligence from infringement to protectability. The classic question is whether the target violates third-party rights; the new question is whether the target's own crown-jewel assets are protectable at all. Where the equity story rests on proprietary content, that question deserves its own workstream.
Deal teams
· Let the warranties reflect thefindings. Where diligence confirms a documented, human-led process, ownership and validity warranties can stand broad. Where it does not, expect knowledge qualifiers on protectability or carve-outs for AI-generated content.
Conclusion
Both the European and the U.S. legislature are expected to provide clarifying rules in the foreseeable future. Until then, the courts have set the direction: copyright follows the human, not the machine. For the M&A market this means that generative AI is no longer only a compliance topic under the AI Act, it goes to the heart of what a buyer actually acquires. W&I insurance can bridge that uncertainty, but as always, only where the risk has been understood, documented and disclosed.
