Assessing risk in the use of AI output as a speed-to-market advantage in creative works
The design industry has always adapted to new tools, but AI has fundamentally altered the design-to-implementation workflow. Historically, software automated repetitive tasks. Today, AI attempts to automate the output itself. Anyone can prompt an AI to generate a logo or vector graphic in seconds. The strategic gap remains in determining whether that output aligns with competitive analysis, market differentiation, and the core brand promise.
Early in my career, I read that creative professionals could do a hack to protect their work by mailing themselves a copy to prove ownership of original work, but according to Copyright.gov, U.S. patent law makes it clear that’s not a legally binding and effective strategy as a substitute to filing for a patent:
"The practice of sending a copy of your own work to yourself is sometimes called a 'poor man’s copyright.' There is no provision in the copyright law regarding any such type of protection, and it is not a substitute for registration."
AI has shattered previous speed-to-market expectations. Teams are moving at the speed of a token budget, fueled by a dangerous overconfidence in systems that promise hyper-optimized workflows.
This rapid acceleration introduces severe business risks. Product leaders face three distinct challenges when evaluating AI-generated work:
• Verifying that the output does not plagiarize existing protected work.
• Ensuring the output is legally clear for commercial use.
• Preventing proprietary workflows and novel service concepts from being fed into public training models.
If your team is rushing to market using AI to generate creative assets, you need to be aware of the concrete legal liabilities defining 2026.
Here are the key legal precedents setting the current standard and how product organizations must adapt to protect their intellectual property.
1. Raw AI output lacks copyright protection.
Case: Thaler v. Perlmutter (U.S. Supreme Court, March 2026) [https://www.hklaw.com/en/insights/publications/2026/03/the-final-word-supreme-court-refuses-to-hear-case-on-ai-authorship]
Precedent: The Supreme Court refused to hear an appeal on AI authorship, cementing the rule that copyright protection requires human authorship.
Recommendation: Mandate rigorous documentation of human intervention, editing, and arrangement for all AI-assisted assets. Unmodified AI outputs default to the public domain, stripping the work of any competitive moat.
2. Prompts and outputs are discoverable.
Case: The New York Times v. OpenAI (U.S. District Court, Ongoing) [https://www.nelsonmullins.com/insights/blogs/corporate-governance-insights/all/from-copyright-case-to-ai-data-crisis-how-the-new-york-times-v-openai-reshapes-companies-data-governance-and-ediscovery-strategy]
Precedent: A May 2025 preservation order compelled OpenAI to retain millions of user conversation logs. Corporate secrecy does not shield infringing use during e-discovery.
Recommendation: Establish strict governance over data inputted into public AI tools. Operate under the assumption that every prompt your team writes is discoverable in a legal deposition.
3. Visual outputs carry strict trademark liability.
Case: Getty Images v. Stability AI (U.K. High Court, November 2025) [https://www.williamfry.com/knowledge/getty-images-v-stability-ai-the-most-important-ai-legal-decision-to-date/]
Precedent: The court held Stability AI liable for trademark infringement for generating outputs containing proprietary watermarks, separate from the legality of the model weights.
Recommendation: Require rigorous output scanning. Prohibit the use of AI to generate recognizable logos, brand identifiers, or assets mimicking proprietary design systems.
4. Global transparency is mandatory.
Regulation: European Union AI Act (Effective August 2025) [https://digital-strategy.ec.europa.eu/en/policies/regulatory-framework-ai]
Precedent: The EU AI Act enforces strict transparency obligations and copyright compliance for General-Purpose AI models.
Recommendation: For products targeting the EU market, mandate the use of enterprise-grade AI vendors compliant with EU data mining opt-outs and explicit content labeling requirements.
Our job as product leaders is not just to innovate. It is to build defensible, legally sound products. Relying on an algorithm to police its own output is a massive blind spot.
