The litigation front against Meta has opened a third front in three weeks.
Publishers and authors filed a class action against Meta and Mark Zuckerberg personally on May 5, alleging willful copyright infringement tied to Llama model training. That case, reported as Elsevier Inc. et al. v. Meta Platforms, Inc. and Mark Zuckerberg, established the theory that a CEO could bear personal liability for training data decisions. A new filing, Hobbs et al. v. Meta Platforms, Inc., advances that theory further. Filed May 22, it reportedly targets individual AI research scientists as named defendants, not just the company’s leadership.
That’s a structural shift.
Prior AI copyright suits named companies. The Zuckerberg suits named an executive. Hobbs, if the reporting holds, names the people who made technical decisions about training data. The legal theory being tested: that researchers who knew about copyright concerns and continued anyway bear personal exposure alongside their employer.What’s confirmed: publishers and authors have sued Meta and Zuckerberg personally for copyright infringement related to Llama training, with court filings and independent reporting alleging Meta used data from shadow libraries including Library Genesis and Z-Library, and that Meta employees were aware of legal questions surrounding that use. The suit is among what legal tracking resources describe as well over 100 active U.S. copyright cases involving AI companies. The scale of active litigation isn’t in dispute. The specific defendant list in Hobbs requires primary court record confirmation.
AI Copyright Litigation, Hobbs v. Meta Positions
Why this matters for compliance teams: the executive liability theory was alarming. The researcher liability theory is operationally significant. If courts allow plaintiffs to name individual engineers and scientists, AI companies face a retention and indemnification question that didn’t exist 60 days ago. Do your employment agreements address litigation exposure for technical decisions? Does your training data documentation account for how sourcing decisions were made and by whom?
The non-obvious implication worth considering: if researcher-level naming survives a motion to dismiss, it creates a disclosure dynamic inside AI teams, people who contributed to training pipelines may start asking questions about indemnification coverage before regulators or plaintiffs do.
The Elsevier filing context is worth keeping: that case was filed in the Southern District of New York and reported as case number 1:26-cv-03689. The Hobbs case venue hasn’t been confirmed from available sources. Both are part of a litigation wave that TJS covered in detail when Zuckerberg was first named personally.
Verification
Partial SVR cross-references (NYT T2, Publishers Association T3, Words and Money T2); primary Wire source was dead Specific defendant names in Hobbs v. Meta not confirmed against court record. Do not publish names without primary source verification. Case venue not confirmed.What to watch
the Hobbs docket, specifically whether the case survives an early motion to dismiss and whether the court allows the individual-scientist theory to proceed. A ruling allowing it to proceed would be a watershed. A dismissal of the individual defendants would narrow the liability theory back to the executive level, but wouldn’t undo the precedent-setting that Elsevier already established. Either way, the federal copyright position on AI training data remains contested, and the courts are currently the primary venue for resolution.
TJS synthesis
Three escalation steps in 18 days, company liability, CEO personal liability, researcher personal liability. Each step has expanded the pool of people with skin in the game. The next question isn’t whether AI companies will face copyright litigation. It’s whether the individuals inside those companies need their own legal strategy.