Q: In one sentence, what’s fundamentally broken about the way AI systems use content today?
A: AI systems ingest and convert content into responses, outputs and decisions and are increasingly trusted with critical workflows – but they operate, by default, without transparency or accountability.
Q: You say last-click attribution was already broken before ChatGPT arrived. Why does the zero-click era finally force the industry to admit it?
A: The attention economy more broadly is in trouble – consumers have been finding ways to avoid being tracked and bombarded with ads for years. Last-click breaks down when consumers’ consideration process takes more than a single step – and full-funnel advertising exists to make sure that consumers are hit from all sides with as many touch points as possible.
Zero-click is actually a misnomer – what is happening is that consumers have found a system that (for now) feels like a safe space to explore ideas and products and solutions, and because it feels like it can aggregate across all content, it can be more neutral. Rather than having to consume all that content and ads and then do the work to determine if this product actually fits their life, they can have that conversation with an agent. Then, once they’ve decided, they go purchase directly.
So the conversation moves from “the last person to get their attention wins” to “brands that can make a compelling case for why their product actually has value wins”.
Q: Models, you have argued, sometimes “err on the side of a lie” rather than risk quoting too much. What does that mean for the accuracy of answers built on publishers’ work?
A: AI systems have almost universally built in guardrails preventing their models from outputting direct quotes from the content they consume in an effort to maintain their claim that they are not substituting that content – they are making something new.
But anytime an LLM “creates” something, there is a chance it gets it wrong. And there is no accountability for those errors in the way that publishers are accountable for incorrect statements, so it isn’t a level playing field.
So we have a system that extracts your work, turns it into something intended to satisfy a user regardless of veracity, and then sends no traffic or payment back while actively using your content to improve their product to the point it doesn’t need you anymore.
Q: SPUR Coalition brings together the BBC, FT, Guardian, Sky and Telegraph around shared technical standards. Why do publishers need a coalition rather than a stack of bilateral licensing deals?
A: The current state of licensing deals supports the few publishers that have enough leverage to get them while doing nothing to set up an environment where publishing and journalism as disciplines can survive. The answer isn’t just getting AI platforms to pay for the content they take (although that is a start!) – we also need mechanisms that allow for understanding how that content is used, accountability for the outputs of these systems, and clear controls for content owners.
No individual deal can move the platform teams to work towards industry wide solutions – that is what coalitions like SPUR can facilitate.
Q: What should a mid-size UK magazine or regional title do this month that does not require waiting for OpenAI or Google to be generous?
A: Block their crawlers, form stronger direct relationships with your audience, move away from “commodity news” and focus on what makes you unique and valuable.
Leaving your content out there and hoping that these systems will somehow start sending traffic back is not a viable strategy – these platforms have made it clear that their goal is to own the information ecosystem. Let’s not continue to freely give them what they need to replace you and hope they change their minds.
Q: What’s next for OpenAttribution, and what would meaningful progress look like over the next 18 months?
A: We’ve got some exciting network integrations in the works that should make it even easier for site owners to use the free tools OA provides and get a better understanding of what their current AI traffic looks like. As the content usage reporting language starts seeing broader adoption, that should allow for a seamless transition from visibility around AI scraping into a real understanding of how often content is used and cited.
Over the next 18 months we hope to see a truly transparent marketplace form between the AI systems that need good inputs and the content producers that create them. AI agents are becoming more sophisticated across organisations – and that reach gives us the opportunity to work with companies that care about safe, reliable, and licensed content usage in their agents. OA will also continue working to help the consumer answer engines consider following a similar path.
About us
OpenAttribution is a neutral, non-profit initiative building open standards and infrastructure for AI content usage, attribution and transparency. As AI systems increasingly retrieve, cite and summarise third-party content, publishers, brands, creators and other content owners need a clearer way to understand how their work is being used.
OpenAttribution is developing open telemetry standards, content owner resolution infrastructure and practical tools that help AI platforms report usage signals back in a consistent, privacy-preserving and interoperable way. The goal is to create a trusted reporting layer for the AI content ecosystem: one that shows when content was retrieved, cited, displayed or engaged with, and enables downstream use cases such as attribution, compliance, verification and fair compensation.
OpenAttribution works with publishers, brands, AI platforms, measurement partners and industry bodies to support a more transparent, accountable and measurable content ecosystem for the AI era.
