The News, Media Alliance (NMA), a trade group representing roughly 2,200 American news publishers, has signed an AI licensing deal with Bria AI, an attribution-focused artificial intelligence company. The goal is to create a recurring revenue stream for small and mid-sized outlets through RAG, or retrieval-augmented generation, licensing.
RAG is how AI systems pull live or stored content from outside sources to build responses, rather than drawing only on pre-trained data. Publishers whose articles feed that retrieval process have historically received nothing for it.
The NMA's deal with Bria is built to change that for its members. Bria's attribution technology tracks which publisher content is accessed and how often, enabling compensation tied to verified usage rather than guesswork.
It is the NMA's second formal AI licensing partnership: the group already operates an agreement with ProRata AI, whose Gist.AI product pays member publishers 50% of revenues generated when their content is used.
Collective Licensing as a Counterweight to Bilateral Big Tech Deals
The logic behind both deals comes down to bargaining power. One small publisher has little pull when negotiating with an AI platform. But 2,200 of them together is a different story, and the NMA's model lets outlets with thin legal and business development budgets plug into licensing infrastructure they could never build on their own.
The Arena Group, a digital media company, has publicly flagged the lack of transparency in AI content licensing deals and the widening divide between large and smaller publishers in securing agreements at all. Big outlets have moved faster and at far greater scale.
Amazon secured a multiyear deal with The New York Times reportedly worth $20 million to $25 million annually, covering the newspaper's news content, NYT Cooking, and The Athletic.
Meta AI has expanded to nine publisher partnerships globally, with a News Corp agreement valued at up to $50 million per year.
Niche Digital Outlets Suffer
Those numbers are simply not on the table for a regional paper or a niche digital outlet running on a fraction of that budget. The NMA and Bria deal does not promise comparable sums. But it offers something the one-on-one model structurally cannot: a seat at the table for publishers who would otherwise be shut out entirely.
Publishers have described the financial impact of AI companies using their content without payment as devastating, with AI-driven traffic erosion piling on top of ad revenue losses that predate the generative AI era. At the same time, 93% of publishers now use AI tools internally, up from 42% in 2022, with roughly half deploying AI only for internal workflows rather than public-facing products. Building with AI while being consumed by it. That is the bind many newsrooms are stuck in.
The Trump administration has declared that training AI systems on copyrighted content does not infringe U.S. copyright law. If that position survives legal challenges, it strips away a core lever publishers had hoped to use to compel payment. Encyclopedia Britannica and Merriam-Webster have filed suit against OpenAI for copyright infringement and trademark dilution, so litigation remains an active front. Still, collective licensing looks like the more productive near-term path for most of the industry.
Across the Atlantic, the Publishers' Licensing Services (PLS) launched what it described as the United Kingdom's first collective AI licensing scheme for publishers of all sizes, offering a parallel model to what the NMA is building in the U.S. Both efforts rest on the same basic premise: individual negotiations with AI platforms will keep favoring the platforms, and pooled licensing is the more durable answer.
The NMA-Bria deal does not specify total payout projections or a launch timeline for disbursements, according to available disclosures. Whether the recurring revenue it promises can meaningfully offset the compounding losses publishers have already absorbed from AI-driven traffic declines is a question the industry will be watching closely.
