On March 2, 2026, the U.S. Supreme Court declined to hear a case challenging the principle that AI-generated works, without human authorship, cannot be copyrighted, according to Fadel. This decision upheld lower court rulings in Thaler v. Perlmutter, solidifying the legal position: content created solely by algorithms lacks federal protection. Creators relying exclusively on generative AI face a significant challenge in safeguarding their output.
Content creators rapidly adopt AI tools, but the legal framework for protecting AI-generated content remains firmly rooted in human authorship. This creates a critical ownership gap, leaving many works vulnerable to uncompensated reuse and infringement. The U.S. Copyright Office confirms that AI-generated content alone cannot receive copyright protection, as detailed by Spencer Fane.
Therefore, companies and individual creators must adapt workflows to ensure demonstrable human contribution in AI-assisted creations. Otherwise, they risk losing exclusive rights and control over their digital assets. While AI platform Terms of Service may grant contractual ownership, this does not guarantee federal copyright protection against third-party infringement; it cannot supersede federal law.
This legal landscape creates a paradox: contractual rights from AI platforms offer no real-world protection against third-party infringement. A platform's Terms of Service (TOS) might state users own generated content, but this is a contract between user and platform, not a federal copyright. This distinction is crucial, as federal copyright provides robust legal standing against unauthorized use. Some AI platforms also reserve rights to reuse or license generated content, especially for free-tier plans, as noted by Fadel. Creators could unknowingly sign away control over their 'work' without gaining federal copyright. Without human authorship, the content is effectively public domain, regardless of platform agreements.
The Dual Challenge: AI's Use of Copyrighted Works and the Expanding DRM Market
In Bartz v. Anthropic PBC, the court ruled that training large language models (LLMs) with copyrighted books constituted fair use, finding the use 'exceedingly transformative'. However, this decision did not grant a blanket exemption for all AI data acquisition. The court denied Anthropic's motion for summary judgment on the plaintiffs' piracy claim, finding that knowingly downloading and copying pirated works was not fair use.
This creates a nuanced distinction for AI developers: training on copyrighted data might be fair use, but pirated data sources can still lead to legal liability. Furthermore, in Andersen v. Stability AI, allegations that Stability acquired copyrighted works, used them to train Stable Diffusion, and incorporated derived images were sufficient to plead direct infringement, according to Spencer Fane. These cases shift the focus from AI training's transformative nature to the ethical sourcing of training data.
The digital rights management market grows due to the increasing need for content protection and secure distribution, according to MarketsandMarkets. This demand now includes complexities from AI-generated content and training data. While AI training may be fair use, acquiring and incorporating copyrighted works into that data can still lead to direct infringement claims. This places the onus on AI developers to ensure ethical data sourcing.
The legal landscape creates a paradox: AI models can legally train on copyrighted works under 'fair use,' yet their outputs are denied copyright protection. This one-way street for intellectual property benefits traditional human creators, whose original works remain unequivocally copyrightable. Conversely, content creators relying exclusively on AI are the primary losers; their works lack federal copyright protection and may be freely used by others without compensation, or even claimed by AI platforms.
Companies encouraging content teams to rely solely on AI are building brands on legally indefensible intellectual property. The U.S. Copyright Office and the Thaler v. Perlmutter ruling confirm AI-generated content lacks copyright protection, exposing companies to significant risk as 'brand assets' could be freely copied. Similarly, creators using free-tier AI services face a double bind: their content lacks copyright protection, and they may unknowingly grant the AI platform broad reuse or licensing rights. This erodes control and commercial potential, turning creative output into a commodity for the platform.
While AI training might be fair use, acquiring and incorporating copyrighted works into training data can still lead to direct infringement claims, a critical shift underscored by the Andersen v. Stability AI case. This places the onus on AI developers to ensure ethical data sourcing and will influence future platform policies. By Q4 2026, many platforms were likely to revise their terms to explicitly address content ownership and data provenance, impacting how creators interact with tools like Midjourney and Stable Diffusion.
