Emily Lyons v. OpenAi Foundation
Issue
Whether this federal court action against OpenAI arising from an AI-linked murder-suicide should be dismissed or stayed under the *Colorado River* abstention doctrine in favor of an earlier-filed, parallel California state court action asserting identical product liability and UCL claims, and separately whether dismissal is required under California Code of Civil Procedure § 377.32 for plaintiff's failure to file the affidavit required of a decedent's successor in interest.
What Happened
Emily Lyons, as representative of Stein-Erik Soelberg's estate, filed this federal action on December 29, 2025, eighteen days after the executor of murder victim Suzanne Adams's estate filed a substantively identical complaint in San Francisco Superior Court; both suits allege that design defects in OpenAI's GPT-4o caused Soelberg to kill his mother and then himself. Defendants OpenAI and Samuel Altman moved to dismiss under Rules 12(b)(1) and 12(b)(6), arguing first that *Colorado River* abstention warrants dismissal because the federal and state actions are parallel — sharing the same defendants, the same seven causes of action, and virtually identical factual allegations — and because the state court has already coordinated the Adams case with more than ten similar ChatGPT product liability suits into a JCCP proceeding. As an independent ground, defendants argue the complaint must be dismissed because Lyons failed to file the affidavit required by California Code of Civil Procedure § 377.32 to establish her authority to sue as the decedent's successor in interest. Defendants do not contest that the claims may be pursued, but argue the proper forum is the state court coordinated proceeding.
Why It Matters
This motion presents an early procedural test of whether federal courts will decline jurisdiction over AI product liability suits in favor of consolidating such claims in state court mass-tort coordination proceedings, potentially channeling the emerging wave of ChatGPT-related personal injury litigation into California's JCCP framework rather than federal court; the outcome may also signal how courts will manage the proliferation of parallel AI liability actions filed by different plaintiffs arising from the same underlying AI-assisted harm.
How accurate was this summary?