Software & Apps

Risks in Figma’s AI design tool copyright exposure for enterprises

Figma faces a class action lawsuit in California following reports that its Make Designs feature produced interfaces nearly identical to Apple's weather app. Enterprise clients face significant copyright infringement and trade secret misappropriation risks when using AI-generated outputs.

Risks in Figma's AI design tool copyright exposure for enterprises

The contractual training risk

Figma faces a proposed class action lawsuit in California federal court filed on November 21, 2025, which alleges the company uses customer design files to train its AI models without consent. This lawsuit includes claims of copyright infringement and trade secret misappropriation. Many design tools include a clause stating the company may use content to "improve our services" within their terms. This language provides the legal basis for training models on user files. It covers confidential product mockups or unreleased brand work. If your client’s confidential product work lives in a tool with this language, the risk of data being used for training remains high. The reported significant drop in Figma’s stock price tied to this news cycle shows that investors recognize these liabilities. The company can use user-generated content to improve its services. Most users do not read the full contract to find these specific permissions. This legal pressure follows the incident where Figma’s "Make Designs" feature produced interfaces nearly identical to Apple’s weather app. This turned an abstract training-data argument into a visible demonstration of risk for all users. Because the "improve our services" clause is often buried in boilerplate, many companies do not realize they have granted such broad licenses to their proprietary assets.

The replica problem

The "Make Designs" feature in Figma, intended to help designers quickly mock up app designs, produced interfaces nearly identical to Apple’s weather app, which forced CEO Dylan Field to disable the tool. This incident, which occurred days after the feature’s rollout at Config 2024 from June 26 to June 28, 2024, demonstrates that AI models can reproduce recognizable, existing products instead of creating original designs. Andy Allen, CEO of Not Boring Software, first identified the issue and warned designers to be cautious of the tool’s lack of originality. Figma uses models from OpenAI and Amazon’s Titan Image Generator G1 instead of training its own models. Figma’s CTO Kris Rasmussen stated that the company does not train the AI models themselves, but the tool’s output still mimicked Apple’s design. CEO Dylan Field attributed the issue to a push for the team to meet deadlines and the use of off-the-shelf AI models alongside a bespoke design system. Because Figma relies on off-the-shelf models, users cannot be certain where the AI’s inspiration comes from. This leads to the risk of unintentional plagiarism. If an AI tool reproduces a specific, recognizable interface, the company using that output faces potential legal claims for infringement. The tool’s output remixes existing designs instead of creating anything genuinely new.

The authorship standard

US copyright law requires human authorship for any work to receive protection. The D.C. Circuit affirmed in March 2025 that the Copyright Office must refuse registrations for works created without human involvement. The US Supreme Court declined to hear the appeal for the Thaler case on March 2, 2026, meaning the lower rulings stand. Purely AI-generated content, such as an image created from a single text prompt, has no legal owner in the US, meaning anyone can copy it without permission. The US Copyright Office published a report in January 2025 stating that text prompts alone do not constitute enough expressive detail to grant copyright to a user who only provides instructions to an AI model. To secure rights, enterprises must show significant human creative input through manual editing or composition. The UK government issued a report in March 2026. It stated that its provision for computer-generated works is unclear and should be removed. In China, the Beijing Internet Court has protected AI-generated images when users make deliberate choices about prompts and parameters. The court found that such creative input meets the standard for originality. Does the current legal framework provide enough protection for companies that heavily rely on AI-assisted workflows? You must document every instance where a human designer modifies, retouches, or arranges AI outputs to build a record of authorship.