Vibe-coding and intellectual property: when AI replicates without copying

What happened?

On June 26, 2026, Marc Seitz, co-founder of Papermark (an open-source platform for data rooms), posted screenshots on X showing that Corgi's new Dataroom product — an insurtech backed by Y Combinator that has raised $374 million from investors like TCV and YC — used exactly the same language and features as Papermark, word for word. According to TechCrunch's report the same day, Seitz detailed that the interface, feature terminology, and user experience were virtually identical. Corgi's CEO and CTO, Nico Laqua, quickly responded on X by posting code 'receipts' to demonstrate that the source code of both products was different. In a statement to TechCrunch, a Corgi spokesperson confirmed that the contested elements 'had already been changed,' but admitted that the AI used in the development process — a practice known as 'vibe-coding' — replicated Papermark's appearance and language without copying its underlying code. This case is not isolated: in 2024, a similar incident with GitHub Copilot sparked controversy when the AI assistant was found to reproduce GPL-licensed code snippets without attribution. However, the difference here is that the copying was not of code, but of design and functional language.

Why is it important?

The Papermark-Corgi case exposes a critical legal loophole in the era of 'vibe-coding,' a term coined to describe AI-assisted programming where the developer relies on automatic generation without fully understanding the output. Legally, copyright protection for source code does not cover the 'look and feel' of an interface, unless it is patented or protected by trade secrets. In the United States, the Lotus v. Borland case (1996) established that menu hierarchies were not copyrightable, but case law has evolved since then. In 2025, a California court accepted a 'trade dress' lawsuit for software, indicating that visual appearance can be protectable if it is distinctive and non-functional. The Corgi case could accelerate this trend: although the code is different, the exact replication of the user experience raises questions about unfair competition and misappropriation of others' efforts. Morally, for the open-source community, the result is indistinguishable from copying, eroding trust in collaborative ecosystems. For startups relying on 'vibe-coding,' the reputational risk is enormous: investors like Y Combinator may demand originality verification clauses, and customers may doubt the product's legitimacy.

Consequences

• Reputational and legal risks: Startups using 'vibe-coding' without human oversight could face lawsuits for unfair competition or trade dress violation, even if the code is original. In the Corgi case, although Laqua claimed the elements were already modified, the damage to Papermark's brand and public perception is done. Legal experts consulted by TechCrunch note that if Papermark decides to sue, it could argue that the deliberate replication of the user experience constitutes misappropriation, especially if it can show that Corgi had access to Papermark's product (which is plausible given it is open source).
• Impact on investors: Y Combinator, which has backed Corgi with $374 million, could face pressure to establish ethical guidelines for AI use in development. In 2025, YC had already updated its terms to include 'code originality' clauses, but this case shows the problem goes beyond. We are likely to see new demands for design and user experience verification, not just code.
• Legal evolution: Courts may begin to consider 'look and feel' as a criterion for infringement. In the European Union, the 2019 Copyright Directive already protects user interfaces as literary works if they are original. In the US, the Corgi case could be the catalyst for the Copyright Office to revise its guidelines. Additionally, the US Patent and Trademark Office (USPTO) may see an increase in design patent applications for software interfaces.
• Data room market: The incident could benefit competitors like DocSend or Firmex, which offer proprietary solutions with clearer legal protections. Papermark, as an open-source project, might see an increase in donations and contributions out of solidarity, but also risks having its work replicated by better-funded startups.

What should readers know?

If you use AI tools to generate code, you must implement a thorough manual review process that verifies not only functionality but also the originality of the interface, terminology, and user experience. Originality no longer resides solely in source code, but in the visual and functional expression of the product. Moreover, current legal protections are insufficient: copyright law does not protect ideas, only expressions, and the line between them is blurry in software. We are likely to see new regulations in the coming years, such as an update to the Digital Millennium Copyright Act (DMCA) to include protections against AI replication, or case law extending trade dress to software interfaces. For developers, the lesson is clear: 'vibe-coding' without legal and ethical quality control is a ticking time bomb. For investors, due diligence must now include an audit of the user experience against competitors. And for the open-source community, this case underscores the need for licenses that protect not only code but also the 'look and feel' of projects, perhaps through 'no interface copying' clauses in licenses like the AGPL.