The relationship between AI systems and news publishers has been extractive by default. Bria is building the infrastructure to change that - and the NMA partnership is where it starts.
Across 81,000 interviews in 159 countries, Anthropic found that unreliability is the single biggest concern people have about AI - above job displacement, above governance, above everything else. The fear isn't abstract: it's the experience of an AI that sounds authoritative but can't show its work.
That gap has a specific cause. AI outputs aren't grounded in sources that users, clients, or regulators would recognize as authoritative. Professional journalism that is current, independently produced, and editorially accountable, is exactly what's missing. The industry that creates this credibility has been cut out of the value it generates.
"Publishers are blocking AI bots because these companies are not willing to pay for the content their model has been trained on, and their output is almost entirely internal."
Telegraph SEO Director - Press Gazette, January 2026Bria's attribution infrastructure tracks exactly how licensed content contributes to AI outputs and routes compensation back to the rights holders whose work made those outputs possible. A single API/MCP server integration gives enterprise AI systems - copilots, research agents, analysis workflows - real-time access to licensed, authoritative content across the news, legal, financial, and scientific domains on which their outputs depend. We've operated this infrastructure at scale for visual content for over two years. The NMA partnership marks the moment professional journalism becomes part of that system.
The key distinction is inference-time access. Content is retrieved as a live, attributable layer at the moment an AI generates a response - not extracted once during model training and absorbed into weights. Every use is logged. Every citation is traceable. Compensation flows back to publishers based on how their content actually performs, not on estimates or flat fees.
Unlike revenue-share or engagement-proxy models, compensation here is tied to exactly how your content contributes to each AI output - tracked at the claim level, not approximated. The more your journalism is relied on to make AI outputs defensible, the more you earn.
No training deal or bilateral agreement tells you which stories are being cited, in which industries, how often. Bria's attribution reporting does. For the first time, publishers have real data on how their journalism functions inside AI systems - and that data is negotiating leverage.
Publishers set the terms: which content is available, which use cases are permitted, which customers or verticals can access it. Existing enterprise relationships stay protected - carve out any customer you already serve directly.
Every NMA member participates on the same commercial terms. Compensation is determined entirely by how often your content is cited - not by the size of your newsroom or the strength of your legal team. A regional paper with authoritative local reporting earns on the same basis as a national masthead.
Professional journalism is the right first vertical: it's current, broadly trusted, and independently produced. The NMA - representing approximately 2,200 news, magazine, and digital media organizations - means we're building this at scale from day one, not publisher by publisher.
For NMA members, participation means joining a licensed content network where your journalism becomes part of the infrastructure that makes enterprise AI defensible - and where that contribution is measured, attributed, and compensated every time it matters.
The network is designed to grow. News is the first vertical. Legal, financial, and scientific content follow. The goal is a market where the value publishers create for AI is continuously recognized - not extracted once and forgotten.
No. This is an additive option, not a replacement. Publishers can and do enter into direct deals with AI companies independently, and participation here doesn't change that. Collective opt-in licenses like the NMA x Bria license offer something direct deals typically can't: scalability across the full membership, lower transaction costs on both sides, and access for publishers who may not have the resources to orchestrate individual negotiations. Bria provides the commercial infrastructure that makes participation viable for the full range of the industry, with the attribution layer that bilateral deals don't always include.
No. Publishers define their own participation terms - including which customers or verticals are excluded from accessing their content through the network. Access policies are enforced programmatically, not managed through trust or contract review.
The structural difference is attribution and control. Traditional aggregators redistribute or display content - publishers are the product, and the value flows to the platform. In Bria's model, content is never exposed to end users directly and is never available for unrestricted download. It functions as a verification and grounding layer inside enterprise AI systems, and every use is tracked and compensated. Publishers also retain ongoing control: over what's available, who can access it, and on what terms.
Yes. Participation is opt-in, which means opt-out is part of the structure. Publishers control their involvement and can adjust or withdraw participation. The NMA's role in this partnership is specifically to ensure that the terms work for its members - if they don't, that's a problem for the partnership, not something publishers absorb.
The infrastructure for how AI systems access and attribute professional content is being built right now. Publishers who participate early help define what fair looks like - the terms, the norms, the commercial model - before those defaults are established without them.
NMA members interested in learning more or participating can reach out through the Alliance or directly to Bria.
Get involved →