Let me frame that differently. Imagine a competitor walked into your office 1,500 times, photographed every document on every desk, went home, built a product using your work, and then sent you a single customer as a thank you.
No Australian business would accept that deal. Yet that is precisely the arrangement every publisher in this country currently operates under with AI companies. And unlike the relationship with Google, there is no existing or proposed legislation that adequately addresses it.
The data is getting worse, not better
TollBit's State of the Bots reports tracked AI scraping activity across its publisher network throughout 2025, and the trajectory is alarming. At the start of the year, there was one AI bot visit for every 200 human visits. By Q4, that had shifted to one bot for every 31 humans, a more than sixfold increase in bot density in under twelve months.
The composition of that scraping is shifting, too. Training crawls actually fell 15% between Q2 and Q4, but retrieval-augmented generation (RAG) bots, the ones that extract real-time information to power AI answers, grew 33% in the same period. AI search indexers surged 59%. The bots are no longer just learning from our content. They are serving it back to users in real time, bypassing the publisher entirely.
And the return traffic? Click-through rates from AI applications dropped from 0.8% in Q2 to just 0.27% in Q4 2025. For publishers with AI content licensing deals, the picture is even bleaker: CTRs collapsed from 8.8% in Q1 to 1.33% in Q4. Even paying for access does not guarantee traffic back. Google still sends 678 visitors for every one referred by an AI application.
Perhaps most concerning: 30% of all AI bot scrapes in Q4 bypassed robots.txt entirely. OpenAI's ChatGPT-User bot accessed content from 42% of sites that had explicitly blocked it. The technical barriers publishers have put in place are being systematically ignored.
The Pew Research Centre confirmed the demand-side impact: when users encounter an AI-generated summary in search results, they click through to the source just 1% of the time. Press Gazette reported that Google search traffic to publishers dropped by a third globally in the year to November 2025. Zero-click searches now account for 69% of all queries. Senior media executives expect a further 43% decline in search traffic over the next three years.
The financial toll across the sector is estimated at approximately $2 billion in annual advertising revenue losses.
But quality publishers still matter (more than ever)
Here is the counterintuitive part of this story. While the aggregate numbers are dire, there is growing evidence that quality publishers are becoming more valuable in the AI era, not less.
An AirOps study found that 85% of brand mentions in AI search results come from third-party sources, not from the brand's own website. Listicles and reviews drive almost 90% of those mentions. Brands that invest in a strong off-site presence are 6.5 times more likely to earn visibility in AI search than through their owned content alone.
What does that mean in practice? It means that when ChatGPT, Perplexity, or Google's AI Mode recommends a product, travel destination, or brand, it is overwhelmingly citing independent publishers who have written authoritative, trustworthy content on those topics. The publisher becomes the reference point. The branded mention has become the new backlink.
For advertisers, this changes the value proposition entirely. A feature in a quality publication is no longer just a traffic play. It is an investment in how your brand appears across an entire ecosystem of AI-powered discovery. If an AI model trusts a publisher enough to cite it, that trust transfers to every brand mentioned within that content.
Google's own head of search, Liz Reid, reinforced this point in a recent interview: "Are you producing great content that is really interesting to people, or are you creating the same junk 100,000 other people are, and just hoping your content is going to surface at the top?"
That is the dividing line. Publishers who churn out the same commodity content that every other news site produces are on a guaranteed path to obsolescence. The ones who add genuine value, bring unique perspectives, conduct original testing, and create content that AI cannot easily replicate will not only survive but also become more important than ever. Human expertise, real-world experience, and editorial judgement are precisely the signals that AI models use to determine which sources to cite and trust.
Australia's regulatory gap
To be fair, Australia has shown more backbone than most jurisdictions. In October 2025, Attorney-General Mark Dreyfus rejected a text-and-data mining exception under the Copyright Act, refusing to let tech companies scrape copyrighted material without compensating creators. The government committed to developing a fair remuneration framework through its Copyright and AI Reference Group.
That was the right call. But the proposed News Bargaining Incentive has a fundamental blind spot. It was designed to regulate platforms that display links to news content. The entire mechanism assumes a link-based value exchange. AI companies don't display links. They ingest content, synthesise it, and deliver answers directly. If legislation only captures platforms that send links, it is regulating yesterday's problem while tomorrow's problem scrapes the web unchecked.
At Man of Many, we submitted our response to the Treasury consultation earlier this year, proposing technology-neutral definitions that capture AI companies, not just traditional search engines. The definition of a "designated digital platform" needs to include any entity that systematically accesses published content and derives commercial value from it, regardless of whether it sends a link back.
What actually needs to happen
First, the News Bargaining Incentive must use technology-neutral language. Any platform that commercially benefits from crawling or ingesting publisher content should be captured, whether it provides links, summaries, or conversational answers.
Second, publishers themselves need to earn their place in this new landscape. The answer is not to demand protection for mediocre content. It is to create work that is genuinely irreplaceable: original reporting, expert analysis, real-world product testing, and perspectives that only come from deep domain expertise. The publishers who do this will find themselves cited more, not less, as AI systems increasingly prioritise authoritative, trustworthy sources.
Third, the forthcoming Productivity Commission final report and Attorney-General's discussion papers on licensing and enforcement need to move from consultation to action. The publishing industry cannot afford another two years of discussion while its content is being systematically extracted.
The window is closing
I am not anti-AI. At Man of Many, we use AI tools daily. We were selected for OpenAI's APAC Newsroom AI Catalyst programme precisely because we are practitioners, not just commentators. We believe in augmentation, not replacement.
But augmentation requires consent. It requires compensation. And it requires a regulatory framework that recognises the difference between a search engine that sends traffic and an AI system that absorbs your content so effectively that the traffic never comes.
The publishers who will thrive are those creating content worth visiting and building a direct community. As Brian Morrissey recently wrote in The Rebooting, the essential shift for publishers is toward "fandoms, not randoms" - moving from a scale approach to genuine engagement. The winners will be those producing unique, expert-led work that is impossible to replicate at scale.
But here is the part the News Bargaining Incentive must account for: AI models are only as good as the content they are trained on and the content they retrieve. If the economics of publishing collapse and quality content stops being produced, there is nothing left to scrape, nothing left to train on, and nothing left to serve back to users. The AI companies know this. The regulation needs to reflect it. Because the 1,500:1 ratio is not sustainable for publishers - and ultimately, it is not sustainable for the AI companies either.
Scott Purcell, CFA, is Co-Founder of Man of Many, Australia's largest independent men's lifestyle publisher. He is a board member of the Digital Publishers Alliance and was named to Campaign Asia-Pacific's 40 Under 40 in 2025.