Intellectual Property

State of play.

• Apple v. OpenAI is the dominant IP litigation of the moment. Apple's federal complaint in the Northern District of California alleges a coordinated scheme — over 400 former Apple employees at OpenAI, candidates encouraged to bring prototypes to interviews, and deception of Apple's manufacturing partners — to misappropriate hardware trade secrets; OpenAI has moved to dismiss and characterizes the suit as "careless" and "oddly personal" (→ Apple sues OpenAI and two ex-employees for stealing trade secrets to build AI hardware, OpenAI seeks to dismiss Apple’s trade-secrets lawsuit over hardware hiring, OpenAI Rejects Apple’s Injunction Bid in Trade-Secret Fight, Apple sues OpenAI, alleging coordinated trade secret theft for AI hardware).
• The AI training data licensing market has formalized. Google, Meta, Microsoft, and OpenAI are shifting from web scraping toward paid licensing agreements — Photobucket's 13-billion-asset archive, paywalled news, platform chat logs — with deal terms largely opaque and pricing ranging from cents per image to higher rates for video .
• The Fastcase v. Alexi dispute is testing whether "internal use" data licenses survive an AI pivot. Fastcase seeks destruction of both the licensed dataset and the AI model weights trained on it, treating model weights as infringing derivative works — a theory that, if adopted, would give data licensors structural leverage over any downstream AI application (→ Fastcase Sues Alexi Over Unauthorized AI Training Using Licensed Case Law).
• The CoStar/CREXi copyright litigation has produced a significant concurrent-conflict disqualification — Quinn Emanuel disqualified for simultaneously representing CREXi while holding a CoStar engagement — with the Ninth Circuit now deciding whether to reverse and whether the district court case proceeds in the meantime (→ CoStar Fights CREXi Bid To Stay Suit Over Quinn Emanuel DQ, CREXi asks 9th Circuit to reverse Quinn Emanuel disqualification in CoStar case).
• For counsel advising technology companies, content owners, or IP-intensive portfolios, the practical baseline is that AI has simultaneously opened three active litigation fronts — trade secret misappropriation via employee mobility, copyright liability for training data, and data license scope disputes — each generating precedent on overlapping timelines with no settled doctrine on any of them.

Where things stand.

• Apple v. OpenAI is the leading trade secret case in AI hardware. The complaint names Tang Tan (former VP of Product Design, now OpenAI's Chief Hardware Officer) and Chang Liu, alleges OpenAI used recruiting to extract confidential iPhone, Apple Watch, and MacBook specifications, and seeks injunctive relief requiring redesign of OpenAI's hardware without Apple's proprietary information; OpenAI's motion to dismiss and Apple's preliminary injunction request are both pending before the Northern District of California (→ Apple sues OpenAI and two ex-employees for stealing trade secrets to build AI hardware, OpenAI seeks to dismiss Apple’s trade-secrets lawsuit over hardware hiring, OpenAI Rejects Apple’s Injunction Bid in Trade-Secret Fight, Apple sues OpenAI, alleging coordinated trade secret theft for AI hardware, Apple asks court to block OpenAI from using alleged trade secrets).
• AI training data fair use remains unsettled but is generating settlement precedent. Anthropic's transformative fair use argument in California and the Bartz v. Anthropic $1.5 billion class settlement are on parallel tracks; the formalization of paid licensing deals by major AI developers signals the market is pricing in copyright risk even before doctrine settles (→ Fastcase Sues Alexi Over Unauthorized AI Training Using Licensed Case Law).
• Data license scope for AI applications is the next major contract battleground. The Fastcase v. Alexi dispute turns on whether an "internal research purposes" restriction in a 2021 data license forecloses a licensee's subsequent pivot to a commercial AI platform — and whether model weights trained on licensed data are themselves infringing derivative works (→ Fastcase Sues Alexi Over Unauthorized AI Training Using Licensed Case Law).
• UK AI platform liability for generated content is being tested in court. British MP Jess Asato has filed a High Court claim against xAI alleging Grok generated sexually explicit fake images of her without consent, invoking the UK Data Protection Act and the tort of misuse of private information; Ofcom has opened a parallel investigation into X over Grok-related imagery (→ UK lawmaker sues xAI to block Grok from making sexualised images).
• Concurrent-conflict disqualification in complex IP litigation carries real strategic weight. The CoStar/CREXi dispute — six years of copyright and antitrust litigation — has been disrupted by Quinn Emanuel's disqualification for simultaneously representing CoStar, with the Ninth Circuit now the arbiter of whether the order stands and whether proceedings are stayed (→ CoStar Fights CREXi Bid To Stay Suit Over Quinn Emanuel DQ, CREXi asks 9th Circuit to reverse Quinn Emanuel disqualification in CoStar case).
• Synthetic performer and digital replica laws are operative at the state level. New York's Fashion Workers Act and synthetic performer disclosure laws require explicit model consent and AI avatar disclaimers; California's AB 2602/AB 1836 operate in parallel; the federal NO FAKES Act remains pending; and the White House preemption push creates a potential collision with state mandates (→ UK lawmaker sues xAI to block Grok from making sexualised images).
• 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 — front-loading RPI arguments before institution is now mandatory strategy .
• Patent disclosure irrevocably forfeits trade secret protection. The Federal Circuit's ruling in International Medical Devices v. Cornell confirms that patent-disclosed designs cannot receive trade secret protection under California's UTSA — a portfolio audit trigger for medtech and any patent-heavy sector running dual-protection strategies .
• 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 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 .

Latest developments.

• OpenAI moved to dismiss Apple's trade secrets complaint, arguing Apple failed to identify any protectable trade secrets or demonstrate plausible misappropriation; a preliminary injunction hearing is pending (→ OpenAI seeks to dismiss Apple’s trade-secrets lawsuit over hardware hiring).
• Apple sought a preliminary injunction to bar OpenAI, io Products, Tang Tan, and Chang Liu from accessing, using, or disclosing alleged Apple trade secrets during litigation; OpenAI publicly characterized the suit as "careless" and "oddly personal" (→ OpenAI Rejects Apple’s Injunction Bid in Trade-Secret Fight, Apple asks court to block OpenAI from using alleged trade secrets).
• Apple's July 10 complaint alleges OpenAI encouraged job candidates to bring Apple prototypes to interviews, deceived Apple's manufacturing partners, and leveraged Tang Tan's insider access before his departure — with over 400 former Apple employees now at OpenAI (→ Apple sues OpenAI and two ex-employees for stealing trade secrets to build AI hardware, Apple sues OpenAI, alleging coordinated trade secret theft for AI hardware).
• Major AI developers are shifting to paid training data licensing — Google, Meta, Microsoft, and OpenAI have moved from web scraping toward proprietary dataset purchases, with deals involving News Corp, Reddit, Amazon, Axel Springer, and The New York Times, per Reuters .
• UK MP Jess Asato filed a High Court claim against xAI seeking damages, a declaration of unlawfulness, and an injunction prohibiting Grok from generating sexualized images of her, invoking the UK Data Protection Act and misuse of private information (→ UK lawmaker sues xAI to block Grok from making sexualised images).
• Fastcase v. Alexi Technologies went to hearing in the District of Columbia — Fastcase seeks destruction of both the licensed dataset and the AI model weights, treating weights as infringing derivative works; Alexi argues its AI pivot was implicitly permitted (→ Fastcase Sues Alexi Over Unauthorized AI Training Using Licensed Case Law).
• CoStar is opposing CREXi's bid to stay the district court proceedings while CREXi appeals Quinn Emanuel's disqualification to the Ninth Circuit; Judge Marshall found Quinn Emanuel violated its duty of loyalty by simultaneously representing CoStar (→ CoStar Fights CREXi Bid To Stay Suit Over Quinn Emanuel DQ).
• CREXi has petitioned the Ninth Circuit to reverse the disqualification order and stay district court proceedings pending appeal; the Supreme Court previously denied CoStar's certiorari petition on the antitrust issues (→ CREXi asks 9th Circuit to reverse Quinn Emanuel disqualification in CoStar case).

Active questions and open splits.

• What constitutes actionable trade secret misappropriation in AI hardware development. Apple v. OpenAI tests whether institutional encouragement of employee mobility — recruiting, interview practices, and supplier access — creates organizational liability for misappropriation, and whether a court will grant injunctive relief that forces hardware redesign before trial; OpenAI's dismissal motion and the preliminary injunction hearing will be the first doctrinal signals (→ OpenAI seeks to dismiss Apple’s trade-secrets lawsuit over hardware hiring, OpenAI Rejects Apple’s Injunction Bid in Trade-Secret Fight, Apple sues OpenAI and two ex-employees for stealing trade secrets to build AI hardware).
• Whether AI training constitutes transformative fair use — and whether the licensing market moots the question. The shift by major AI developers to paid licensing deals signals commercial acknowledgment of copyright risk even as the fair use argument remains pending in court; the Bartz v. Anthropic settlement and Anthropic's California filing are on parallel tracks that could produce contradictory signals (→ Fastcase Sues Alexi Over Unauthorized AI Training Using Licensed Case Law).
• Whether model weights trained on licensed data are infringing derivative works. Fastcase's theory — that the model itself, not just the training run, is the infringing artifact — is untested; if adopted, it would give data licensors structural leverage to demand model destruction as a remedy, reshaping every AI data license negotiation (→ Fastcase Sues Alexi Over Unauthorized AI Training Using Licensed Case Law).
• Whether AI platforms bear direct liability for harmful content their systems generate. The Asato v. xAI claim in the UK High Court tests whether a platform's liability attaches at the system level or only to individual users — a question with direct implications for how AI companies structure content moderation obligations and indemnification in deployment agreements (→ UK lawmaker sues xAI to block Grok from making sexualised images).
• Concurrent-conflict rules when large firms hold opposing-party engagements in separate matters. The CoStar/CREXi disqualification — Quinn Emanuel representing CREXi while holding a CoStar engagement in a separate federal matter — will clarify how the Ninth Circuit applies concurrent-conflict doctrine in multi-matter firm relationships, a recurring issue for firms with large commercial IP practices (→ CoStar Fights CREXi Bid To Stay Suit Over Quinn Emanuel DQ, CREXi asks 9th Circuit to reverse Quinn Emanuel disqualification in CoStar case).
• 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 develops more rigorous pre-institution procedures in response .
• 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, and the xAI UK litigation adds a cross-border dimension to platform liability for synthetic content (→ UK lawmaker sues xAI to block Grok from making sexualised images).

What to watch.

• The Northern District of California's ruling on Apple's preliminary injunction against OpenAI — if granted, it will materially constrain OpenAI's hardware timeline and signal courts' willingness to impose pre-trial restrictions in AI trade secret disputes; if denied, it shifts the battleground to the dismissal motion.
• The D.C. district court's ruling in Fastcase v. Alexi on whether "internal use" data license restrictions survive an AI pivot and whether model weights constitute infringing derivative works — the decision will reshape how AI data licenses are drafted and enforced across the industry.
• The Ninth Circuit's disposition of CREXi's petition to reverse Quinn Emanuel's disqualification — and whether the court grants a stay of district court proceedings pending appeal, which would determine whether CREXi must find and prepare new counsel under trial pressure.
• Whether the Asato v. xAI High Court claim produces interim injunctive relief against Grok — the first judicial order directly restricting an AI system's generative output in a privacy/data protection context would set a significant cross-border precedent.
• The trajectory of paid AI training data licensing deals — whether terms become publicly disclosed through litigation or regulatory disclosure requirements, which would establish market-rate benchmarks for every pending copyright negotiation.
• Whether Congress formally advances the NO FAKES Act or the Trump America AI Act in a way that addresses copyright liability for AI training data, which would preempt the pending fair use litigation and reshape the licensing market simultaneously.