What These Deals Establish
By April 2026 the bilateral AI content licensing layer has matured into a recognisable pattern: large-publisher / large-AI-lab agreements covering training-data rights, real-time data feeds, attribution requirements, and increasingly explicit per-use pricing. The deals are not the dominant volume in the AI content economy (the marketplace layer carries far more transactions), but they set the upper bound on per-content pricing and establish the contractual norms that smaller agreements increasingly imitate.
This page is a reference catalogue of the major publicly disclosed bilateral deals as of April 2026, with the recurring patterns that brand and publisher teams should recognise when evaluating their own positioning.
Major Disclosed Deals
| Publisher | AI Lab | Year | Reported scope |
|---|---|---|---|
| News Corp | OpenAI | 2024 | Multi-year, multi-property; training plus real-time feeds |
| Axel Springer | OpenAI | 2023 | Multi-year; ChatGPT integration plus training data |
| Le Monde | OpenAI | 2024 | Multi-year; French-language coverage focus |
| Vox Media | OpenAI | 2024 | Multi-year; multi-property training plus product integration |
| The Atlantic | OpenAI | 2024 | Strategic content partnership |
| News Corp Australia | OpenAI | 2024 | Subset of News Corp parent deal scoped regionally |
| FT (Financial Times) | OpenAI | 2024 | Multi-year; ChatGPT integration |
| AP (Associated Press) | OpenAI | 2023 | Two-year, training-focused |
| 2024 | Reported $60M/year; broad content access | ||
| OpenAI | 2024 | Strategic partnership including ChatGPT integration | |
| Reuters | Meta | 2025 | Multi-year; news content for Meta AI |
| NYT | OpenAI / Microsoft | Litigation pending | NYT lawsuit ongoing; representative of unresolved high-stakes positioning |
| Wiley | Multiple AI labs | 2024-2025 | Academic content licensing |
| Taylor & Francis (Informa) | Microsoft | 2024 | $10M+; academic content |
Recurring Patterns
Six patterns recur across the publicly disclosed deals.
Multi-year scope. Deal terms typically run 2 to 5 years, with extension options. Single-year deals are rare because the operational integration cost on both sides justifies longer commitments.
Bundled training and real-time access. Most deals cover both training-data rights (the lab can use the content to train models) and real-time data feeds (the lab can fetch live content for inference-time citation). Splitting these is operationally clean but reduces the publisher's leverage.
Product-integration components. Many deals include AI-product-integration commitments (the lab integrates the publisher's content into a specific product surface, e.g., ChatGPT showing FT articles on relevant queries). This is meaningful because it converts the licensing fee into a visibility benefit alongside the cash payment.
Attribution requirements. Deal terms increasingly include attribution requirements specifying how the publisher must be credited when content is cited. This is one of the layers ai.txt and ERC-8004 are positioning to standardise.
Exclusivity and territoriality. Some deals carry partial exclusivity (the publisher cannot license the same content to specific competing AI labs for the deal term) or territorial scoping (the deal covers specific markets only).
Implied per-citation rates significantly higher than marketplace. When publicly disclosed total deal values are divided by published or estimated cited-volume figures, the implied per-citation rate is meaningfully higher than marketplace rates. This is partly because the bilateral deal carries fixed-fee components for training rights and integration that marketplace per-fetch rates do not include.
What Smaller Publishers Should Take From This
Most publishers will not negotiate bilateral deals with AI labs because they are too small to attract BD attention. The patterns above still matter for two reasons. First, marketplace contract terms imitate bilateral structures over time, so the patterns indicate where the marketplace layer is heading. Second, when bilateral discussions become possible (typically through publisher-association or agency representation), the patterns are the reference points for negotiation.
Specifically, smaller publishers should expect bilateral discussions to involve multi-year terms, training plus real-time bundling, attribution requirements, and pricing anchored against marketplace rates with a multiplier reflecting the bundle plus the certainty premium. The certainty premium for bilateral deals over marketplace participation typically ranges from 2x to 10x at the per-citation level.
Methodology
Deals listed are publicly disclosed through company announcements, regulatory filings, or news reporting. Implied per-citation rates are calculated from publicly disclosed total deal values divided by published or estimated cited-volume figures. Patterns are derived from analysis of disclosed deal structures across the catalogue. April 2026 point-in-time; the catalogue updates as new deals are disclosed.