News/Media Alliance Partners with Bria AI to Unlock RAG Revenue for 2,200 Small and Mid-Sized Publishers
The News/Media Alliance has partnered with Bria AI to offer its 2,200 member publishers an opt-in AI licensing agreement covering retrieval-augmented generation (RAG) use cases. Under the deal, publishers choose which content to license; Bria aggregates it and distributes it to enterprise AI clients — including copilot builders, legal AI platforms, and financial services AI tools — that use it to ground their responses. Revenue is split 50/50 between Bria and each publisher based on an attribution model tracking how often their content powers AI outputs. The agreement is non-exclusive and uses a templated contract developed jointly by NMA and Bria. NMA also holds a similar deal with AI startup ProRata. The partnership is designed to give smaller publishers — local news outlets, niche titles — access to AI licensing revenue that previously required the scale of publishers like the Financial Times, AP, or The Economist to negotiate directly.
Analysis
The News/Media Alliance's partnership with Bria AI is the most practically significant AI licensing development for small and mid-sized publishers since the first direct deals between major outlets and AI companies began in 2023 — not because of the money it will generate immediately, but because of the infrastructure it creates.
The core problem the NMA–Bria deal addresses is one of negotiating leverage. When OpenAI, Google, or Microsoft wants to license news content, they negotiate with publishers who have the scale and legal resources to sit across the table from a trillion-dollar company. The Financial Times, AP, and The Economist can do this. A regional newspaper with a staff of twelve cannot. The NMA–Bria model inverts this dynamic by aggregating the content of 2,200 publishers into a single licensed pool, giving enterprise AI clients a single point of access to a breadth of vetted, factual content that no individual small publisher could offer.
The RAG focus is strategically astute. The consumer LLM market — ChatGPT, Gemini, Claude — has attracted most of the public attention around AI licensing, but the enterprise RAG market is arguably larger and more durable. More than half of enterprises are now deploying internal AI agents that need to retrieve current, accurate, domain-specific information to answer employee queries. A financial services copilot that surfaces news coverage of regulatory developments, or a legal AI tool that cites recent court reporting, needs exactly the kind of content that NMA's member publishers produce. The demand is real, it is growing, and it is not dependent on the outcome of any copyright lawsuit.
The 50/50 revenue split and the attribution model are the elements that will determine whether this deal delivers meaningful value to publishers. Bria's attribution model — tracking how often a specific publisher's content contributes to a specific AI output — is technically complex and, at this stage, unproven at scale. The comparable ProRata model has been operating longer and has attracted publishers including the Guardian, but the revenue per publisher has not been publicly disclosed. Until NMA members can see actual payment statements, the deal's value is theoretical.
The templated contract is the most durable element of the partnership. By creating a standardised AI licensing agreement that any NMA member can use, the Alliance is building the legal infrastructure for a collective licensing market that does not currently exist for news content. This is the model that the White House's National AI Policy Framework called for when it recommended Congress create a collective licensing system for AI training — and NMA is building it without waiting for Congress to act.