AI Content Disputes
& Infringement Defense

AI content infringement covers two main scenarios. First, AI-generated output that copies existing copyrighted, trademarked, or right-of-publicity-protected work. Second, AI training that ingested copyrighted material without authorization. The USCO Part 2 Copyrightability Report and the USCO Part 3 Generative AI Training Report frame how regulators currently treat each scenario. The legal frame depends on which element was misused: copyright statute for scraping or output similarity, trademark for brand cloning, right of publicity for voice or likeness, and DMCA mechanics for takedown enforcement on user-generated platforms.

Pre-litigation moves are available, and the right ones depend on what you can prove and what you want. The Bartz v. Anthropic class settlement and the Authors Guild path show the demand-letter and group-claim mechanics that have moved the market. SGL handles demand letters to the AI company, settlement-track correspondence, opt-out enforcement guided by the USCO Part 3 report, and coordinated investigation across AI tools that may have ingested your work. Court filing is referred to outside litigation counsel if escalation occurs.

DMCA takedowns are the primary tool when copyright applies, and platform-specific impersonation reports cover the rest. Under 17 U.S.C. § 512, platforms must respond to a properly-formatted notice within statutory windows. AI-generated output that copies your copyrighted work qualifies for §512 the same way human-generated infringement does. The same incident usually appears across three or more platforms, so coordinated attorney-filed takedowns move faster than single-victim filings. Where copyright does not apply, demand letters under right of publicity, the Lanham Act, or platform impersonation policies cover the gap.

The right defense move depends on whether your use is fair use, whether the AI tool's training data is the source, and what indemnification your AI tool terms provide. The four-factor analysis under 17 U.S.C. § 107 governs fair use across purpose, nature, amount, and market effect. Most consumer AI tool terms place indemnification on the user, meaning claims against you do not route back to the AI provider. SGL handles the written opinion, the demand-letter response, the DMCA counter-notice if applicable, and the settlement-track correspondence.

Fair use can be a defense, but the analysis is fact-specific and the bar is higher when your input source is an AI trained on copyrighted material. The four factors at 17 U.S.C. § 107 cover purpose, nature, amount, and market effect, and transformation is heavily weighted in current case law. The Lenz v. Universal Music good-faith standard requires copyright holders to consider fair use before sending a takedown, giving you leverage in the response. A written fair-use opinion supports your counter-notice, demand-letter response, or settlement position.

The USCO Part 2 Copyrightability Report addresses AI output, and the USCO Part 3 Generative AI Training Report (Pre-Publication) addresses training data ingestion. Part 2 holds that purely AI-generated work receives no copyright protection, while AI-assisted work where a human directs, curates, edits, or arranges the output can receive protection where the human contribution is sufficient. Part 3 holds that ingestion of copyrighted works without authorization or licensing raises substantive copyright issues that may require licensing markets to resolve. Both reports inform every demand letter, opinion, and counter-notice that SGL drafts in this category.

Generally no, under 47 U.S.C. § 230, which gives platforms broad immunity for user-generated content. The immunity has limits: Section 230 does not protect platforms from federal criminal liability or intellectual property violations, which is where DMCA §512 mechanics apply. Platforms that follow §512 safe-harbor procedures are insulated from secondary copyright liability. Most major platforms also enforce voluntary policies against misleading AI content. The practical answer is that platforms remove reported AI infringement under their own terms of service, even where Section 230 shields them from formal legal liability.

Multiple AI copyright cases are active or recently resolved, and the legal landscape moves quarterly. The Bartz v. Anthropic class action settled at $1.5 billion, the largest AI copyright settlement on record. The New York Times v. Microsoft / OpenAI case continues to develop. The USCO Part 3 Generative AI Training Report frames the current regulatory analysis. Each demand letter, opinion, and settlement-track move SGL drafts in this category is tied to current authority, and the FAQ section is reviewed quarterly.

It's both, and the right SGL page depends on which question you are answering. The AI Copyright & Authorship page handles the upstream copyright analysis on AI-assisted work you make today: what you can protect, how to document your contributions, and how to register. This page handles the dispute response on the older portfolio that was scraped: demand letters to the AI company, opt-out enforcement, and settlement-track correspondence. The two pages route in sequence: AI Copyright & Authorship answers what you own, this page enforces against unauthorized AI use of what you own.

Both can apply, and the choice depends on the underlying claim. The DMCA Counter-Notice & Demand Letter Response page handles a single §512(g) counter-notice mechanic: challenging the platform's removal directly. This page handles the broader dispute envelope: multi-platform takedowns, demand-letter exchanges, training-data exposure analysis, and AI tool indemnification posture. A single counter-notice often sits inside a larger AI dispute; the strategy call sets which page actually engages.