Three steps to save publishers from AI
Opinion
Working through these steps will not be easy and will be vigorously opposed by AI businesses, which will claim this is an attempt to stifle innovation. But this approach isn’t anti-AI, it’s anti-theft, says the co-founder of the Movement for an Open Web.
At a recent event on AI and Media Plurality in London, the scale of the challenges facing publishers from the rise of AI was laid bare.
Publishers – particularly independents and specialists – are seeing steep drops in traffic and, therefore, revenue as Google and others plunder their content and steer clicks towards AI-generated outputs.
Attendees from the publishing industry described the situation as ‘dire’ and talked of many publishers who’ve already failed under AI-induced traffic loss, with many more facing extinction during 2026.
This isn’t just bad news for independent publishers. Media plurality is central to an effective society. We’ve seen all too graphically how misinformation can poison the body politic, and the death of swathes of independent media voices will only exacerbate this problem.
The issue is, at its heart, that AI businesses rely on publishers for the content which feeds their products, but do not share the fruits of any subsequent monetisation with the sources of that content.
AI businesses are free-riding on publishers’ creative efforts to build answer machines that steal traffic on which those businesses rely.
This is not sustainable. Without fair recompense for the use of their content, publishers will cease to exist, to the harm of consumers and, one assumes, to the AIs, which will no longer have access to quality training data.
However, amidst the gloom, there are glimmers of light. Whilst there may not be 100% consensus on the details, the beginnings of a structure are emerging, which could lead to equitable co-existence between publishers and AI.
Unbundle
In the short term, the biggest damage to content owners isn’t from AI as a whole; it’s from Google’s AI Overviews (AIOs), which siphon traffic from search.
Google’s AIOs are by far the biggest game in town when it comes to traffic – and therefore publisher harm – and that’s because they are leveraging Google’s monopoly position in search.
The issue is that Google is using its dominance in search to promote its new AI products by pushing them to the top of its Search Engine Results Pages (SERPs), thereby capturing a disproportionate share of traffic that would otherwise go to publishers. Similarly, they combine AIO results in Android and many other products.
The only solution to this is unbundling – the removal of AIOs from Google’s Search, Discover and other products to enable publishers and other AI companies to compete on a level playing field.
AIOs could continue to exist as a separate product that can compete alongside other AIs, but their placement at the heart of Google’s primary consumer interfaces is causing untold damage.
This isn’t a new approach. The Microsoft monopoly trials of the 90s focused on its bundling practices, and the remedies resulted in an enforced unbundling that, ironically, led to the rise of the likes of Google.
In the short term, unbundling AIOs would restore traffic to independents and enable them to survive long enough to be part of a sustainable long-term solution.
Whilst this may feel separate from the core long-term challenge that AI poses to publishers, it is an essential step. This is a monopoly problem, not an AI problem, but solving it quickly is essential to the long term health of publishers and the development of AI.
Measure and control
Once the immediate threat posed by Google AIOs is removed, the next step will be to fully understand content rights and usage.
Currently, AIs operate as black boxes, giving publishers little to no control over whether their content is harvested and no understanding of what happens to it once it’s processed.
For example, Google insists on using a single harvester for search and AI, meaning that anyone opting out of AI will lose any search placement.
Effective, granular, and non-discriminatory opt-in controls over the use of publisher content, and enforced transparency about how content is used in processing and user answers, will help publishers regain control over their content and understand the likely value of their content to AI businesses.
With a granular approach to content harvesting and subsequent use surfaced at the point of consumption, the underpinning technical standards will be in place to support payment for derivative works and provide people with assurances concerning the sources of information they might rely upon.
Compensate
Once the nature of AI content use is better understood, a debate is needed about different compensation models for publishers.
There are many ways in which this could happen, from traditional content licensing deals to payments for derivative works to a central cultural fund based on an advertiser levy.
Whatever way is chosen, effective structures to address compensation and unjust enrichment are needed to preserve cultural assets and the long-term sustainability of web and AI projects.
AI companies may argue that they have a unique approach that doesn’t invoke copyright concerns, but the fact is that their systems currently steal publishers’ content to provide a service that they are monetising (or hope to monetise).
The ethical requirement on them to recompense the original content provider is unarguable. This must be made legally binding.
Working through these stages will not be easy and will be vigorously opposed by AI businesses, which will claim this is an attempt to stifle innovation. But this approach isn’t anti-AI, it’s anti-theft.
AI is an exciting technology that could change much in our world – but it can’t be built for the long term if its growth is predicated on the strip mining of publishers.
For a sustainable future for publishers, AI, and society, a fair approach to content use and compensation needs to be put in place now.
James Rosewell is the co-founder of the Movement for an Open Web (MOW)