Submit incident
Documented

A $446 Million Lawsuit Says Meta Pirated Nearly 3,000 Films to Train Its AI Video Generator

September 8, 2026
Curated by Team Raidu · Reviewed by Shiva Ganesh
oecd:2026-09-08-e710View source ↗
LinkedInX

What happened

Strike 3 Holdings filed a $446 million lawsuit against Meta, alleging the company illegally downloaded nearly 3,000 copyrighted adult films and used them to train visual generative AI models. The filing names Movie Gen, Meta's flagship video generation system, as one of the products built on that allegedly stolen material.

The complaint centers on unauthorized reproduction at scale. Strike 3 Holdings, which produces and distributes adult films, claims Meta acquired the content without a license and without payment, using it as training data for systems designed to generate photorealistic video. The volume alleged, close to 3,000 titles, is not an accidental inclusion in a scraped dataset. It implies a deliberate acquisition effort directed at a single rights holder's catalog.

What moves this beyond a standard copyright dispute is the allegation that a Meta executive was directly involved in the unauthorized acquisition. The record does not identify the individual or specify their title, but the implication is that the conduct was not a rogue scraping operation running below management awareness. Someone with authority over AI development resources either ordered it or knew about it and did not stop it.

AI companies have faced recurring scrutiny over training data provenance, with courts still sorting out whether ingesting copyrighted material constitutes infringement, fair use, or something the law has not yet caught up with. What distinguishes this case is the alleged specificity of the violation. Downloading nearly 3,000 individual titles from one rights holder looks less like the broad web-crawl defenses that have carried weight in other litigation, and more like a targeted collection effort with a paper trail attached to it.

The accountability gap here is not just about whether Meta had permission. It is about whether anyone inside the company maintained a clear, auditable record of what data was gathered, who authorized each acquisition, and what the rights status of each asset was at the time it was used. A provable record of what a system was trained on, and who approved each data source, would not prevent a lawsuit, but it would make the internal decision trail visible from the moment a claim was filed rather than leaving it to discovery. When that record does not exist, the first version of events belongs entirely to the company facing the complaint.

Reported impact

Affected parties
Not publicly disclosed
Harm type
Not publicly disclosed
Scale
Not publicly disclosed
Financial impact
Not publicly disclosed
Regulatory action
Not publicly disclosed

Classification

Organization
Not publicly disclosed
AI system
Not publicly disclosed
Industry
Not publicly disclosed
Country
Not publicly disclosed
Provider
Not publicly disclosed
Incident type
Not publicly disclosed

Relevant governance controls

Governance control mapping is not available for this record.

  • No controls mappedNot publicly disclosed

Control mapping is analytical. It does not state that any control would have prevented the incident.

Sources and evidence

This record was researched and written by the Index. The event is also catalogued in the following database, which is listed for cross-reference.

OECD AI Incidents Monitor
Also catalogued in
A $446 Million Lawsuit Says Meta Pirated Nearly 3,000 Films to Train Its AI Video Generator
2026-09-08