Intellectual Property

State of play.

• AI-generated likeness and right-of-publicity enforcement is accelerating on multiple fronts. New York's synthetic performer consent laws take effect June 19, 2026, Dua Lipa has sued Samsung for $15 million over unauthorized image use in TV packaging, and the NO FAKES Act remains pending federally — creating a compliance crunch for brands and agencies operating across jurisdictions .
• AI training data copyright is at a doctrinal inflection point. Anthropic has argued transformative fair use for Claude's training in California federal court, while the $1.5 billion Bartz v. Anthropic class settlement — covering over 100,000 rights holders — moves toward a fairness hearing, with the outcome likely to set a damages benchmark for the field .
• The Federal Circuit has closed the appellate door on PTAB real-party-in-interest challenges, ruling in FedEx v. Qualcomm that RPI objections are integral to the institution decision and unreviewable under § 314(d) — a structural shift in IPR strategy that requires front-loading RPI arguments before institution .
• Cross-border AI-IP divergence is now a compliance planning problem, not a theoretical one. China, the UK/EU, and the US apply materially different standards to AI-generated outputs — with the EU AI Act binding in August 2026 — and no jurisdiction has produced settled guidance on agentic AI ownership .
• For counsel advising technology companies, entertainment clients, or patent-heavy portfolios, the practical baseline is that AI is simultaneously reshaping the tools of IP practice (USPTO's new image search), the substance of IP doctrine (fair use for training data, synthetic performer consent), and the procedural landscape (PTAB finality), all on overlapping timelines.

Where things stand.

• AI training data fair use remains unsettled but is generating settlement precedent. Anthropic's transformative fair use argument in California and the Bartz class settlement are the two most immediate data points; approval of the settlement could establish a damages floor that shapes every subsequent AI copyright negotiation .
• Synthetic performer and digital replica laws are now operative at the state level. New York's Fashion Workers Act and synthetic performer disclosure laws (effective June 19, 2026) require explicit model consent and AI avatar disclaimers; California's AB 2602/AB 1836 operate in parallel; the federal NO FAKES Act is pending; and the White House's preemption push creates a potential collision with state mandates .
• Right-of-publicity and Lanham Act enforcement is active in federal court. The Dua Lipa v. Samsung complaint — alleging copyright infringement, trademark infringement, right of publicity violations, and false endorsement over a manipulated backstage photograph used on TV packaging — tests the outer boundary of image rights in mass-market consumer goods and the damages exposure when a cease-and-desist is ignored .
• PTAB institution finality has hardened. The Federal Circuit's precedential ruling in FedEx v. Qualcomm forecloses appellate review of RPI determinations made at the institution stage, eliminating a procedural avenue petitioners had used to challenge unfavorable outcomes .
• Patent disclosure irrevocably forfeits trade secret protection. The Federal Circuit's ruling in International Medical Devices v. Cornell confirms that dual patent-and-trade-secret strategies on the same subject matter create litigation vulnerability — a portfolio audit trigger for medtech and any patent-heavy sector .
• AI-generated code ownership is unresolved. Autonomous coding agents capable of producing production-ready software from user specifications are deployed commercially, but no court or regulator has determined whether ownership vests in the user, the AI developer, or neither .
• Cross-border AI-IP divergence requires jurisdiction-specific governance. China protects AI outputs with meaningful human input; the UK and EU require human authorship; the US relies on human contribution and fair use — and the EU AI Act's August 2026 binding date adds a hard compliance deadline for multinationals .
• USPTO has deployed AI tools across trademark examination. The beta AI-powered image search tool, a mark description generator, and the Class ACT classification automation tool are now available — supplementing but not replacing comprehensive clearance searches, particularly for design-heavy applications .
• Federal legislative pressure is building but unresolved. The AI Foundation Model Transparency Act (H.R. 8094) imposes disclosure-only obligations on LLM developers; the Trump America AI Act discussion draft would codify federal preemption of state AI laws; over 600 state AI bills were introduced in Q1 2026 alone .
• The Musk v. OpenAI trial is testing whether founder commitments are legally enforceable. Musk's testimony that OpenAI's nonprofit-to-for-profit conversion constituted breach of foundational commitments — with OpenAI valued at $852 billion — raises questions about the enforceability of informal founding agreements in high-stakes tech ventures .

Latest developments.

• Dua Lipa filed a $15 million federal complaint against Samsung alleging copyright infringement, right of publicity violations, and false endorsement over a manipulated backstage photograph used on TV packaging after Samsung dismissed a cease-and-desist .
• Federal Circuit issued a precedential ruling in FedEx v. Qualcomm holding that PTAB real-party-in-interest determinations under § 312(a)(2) are unreviewable on appeal under § 314(d), closing a procedural avenue for petitioners .
• New York's Fashion Workers Act and synthetic performer disclosure laws take effect June 19, 2026, requiring explicit model consent for digital replication and AI avatar disclaimers in advertising .
• Venable's cross-border AI-IP analysis maps the China/UK/EU/US divergence on AI-generated output protection and identifies agentic AI ownership as the most significant unresolved gap across all jurisdictions .
• Anthropic argued transformative fair use for Claude's training in California federal court; the Bartz v. Anthropic $1.5 billion class settlement fairness hearing is scheduled for May 14, 2026 .
• Autonomous AI coding agents generating production-ready software have triggered practitioner debate over IP ownership allocation — user, AI developer, or neither — with no settled doctrine .
• Federal Circuit reversed in International Medical Devices v. Cornell, holding that patent-disclosed designs cannot receive trade secret protection under California's UTSA — a portfolio audit trigger for dual-protection strategies .
• Alston & Bird's April 2026 AI Quarterly documents the legislative pressure: H.R. 8094 (AI Foundation Model Transparency Act), the Trump America AI Act discussion draft, and 600+ state AI bills in Q1 2026 .
• Musk testified at trial that OpenAI's nonprofit-to-for-profit conversion constituted breach of foundational commitments; OpenAI disputes any binding obligation existed .
• USPTO launched a beta AI-powered image search tool for trademark clearance alongside a mark description generator and the Class ACT classification automation tool .

Active questions and open splits.

• Whether AI training constitutes transformative fair use. Anthropic's California filing and the Bartz settlement are on parallel tracks — a court ruling on the fair use argument and a settlement approval could produce contradictory signals on the same conduct within weeks of each other .
• Federal preemption vs. state synthetic performer laws. New York and California have enacted consent-based digital replica regimes; the White House preemption push and the pending NO FAKES Act create a collision course — brands complying with state law today may face a different federal standard by year-end .
• Who owns AI-generated code. No court has resolved whether production-ready software generated by autonomous agents vests ownership in the user, the AI developer, or falls into the public domain — making IP representations in M&A and licensing transactions a live exposure .
• RPI front-loading as IPR strategy imperative. FedEx v. Qualcomm forecloses appellate correction of PTAB RPI determinations — the open question is how aggressively patent owners will now press RPI objections pre-institution and whether PTAB will develop more rigorous pre-institution RPI procedures in response .
• Patent-vs.-trade-secret dual protection viability. International Medical Devices v. Cornell confirms that patent disclosure forfeits trade secret status under California's UTSA — but the split on trade secret identification requirements under the DTSA in federal courts remains active, and the interaction between federal DTSA standards and state UTSA applications is unsettled .
• Cross-border AI-IP governance gap for agentic systems. China, UK/EU, and US apply materially different standards to AI-generated outputs; no jurisdiction has produced guidance on agentic AI specifically; and the EU AI Act's August 2026 binding date creates a hard compliance deadline that may force multinationals to implement governance frameworks before doctrine is settled .
• Lanham Act false endorsement scope in digital manipulation cases. The Dua Lipa complaint tests whether manipulating and redistributing a celebrity's owned photograph in product packaging — without any licensing relationship — constitutes actionable false endorsement, and what damages theory (profits, statutory, or actual) governs when a cease-and-desist is ignored .

What to watch.

• The May 14, 2026 Bartz v. Anthropic fairness hearing — approval or rejection will set the damages reference point for AI copyright settlements and signal how courts will treat the fair use argument in parallel litigation.
• New York's June 19, 2026 effective date for synthetic performer consent laws — expect enforcement guidance from the Department of Labor and early compliance disputes from brands that have already deployed AI avatars.
• The EU AI Act's August 2026 binding date — the first binding multilateral AI regulation, with €15 million penalty exposure for non-compliant AI-altered content labeling.
• Samsung's response in the Dua Lipa litigation — the defense theory (third-party licensing, fair use, or consent) will define the litigation posture for celebrity image manipulation cases in the consumer electronics sector.
• Whether Congress formally introduces the Trump America AI Act or advances H.R. 8094, and whether either bill addresses copyright liability for AI training data in a way that preempts the pending fair use litigation.
• PTAB's procedural response to FedEx v. Qualcomm — whether the board develops more rigorous pre-institution RPI procedures now that appellate correction is foreclosed.