The bot bookshop, breaking news habits are breaking, and the CMA opens the app store gate

Tidal bets on people, Discover card ate everyone, infinite scroll back in the dock, and hallucinations breaking a business - all in this week's Content Aware.

Published: 15:26, 02 July 2026

Corbidge comments on...the art gallery analogy
Google's top legal mind, Kent Walker, has published a white paper that compares AI training on online content to "an art student taking inspiration from walking through a gallery". Our resident Analogy Inspector Rob points out that he has seen this before, almost word for word,
in The Tony Blair Institute's paper from April 2025. Couldn't they come up with something better? Or did they simply ask Gemini for help? The analogy doesn't really hold: inspiration is, after all, a complex human process, LLM training is the taking of someone's original work to produce an amalgam of it. The war is still on, with over a 100 court cases running in the US and both sides lobbying for their cause ferociously.
Read

Publishing & Media

Is the news instinct weakening? 
Pew Research asked 3,560 Americans where they go to for breaking news updates, and habits are changing. Around 36% said they go to their preferred organisation, which is down from 54% a few years ago. According to the study, search engines are chosen by 28%, social media 19%, and AI chatbots only 1% - albeit that 1% possibly does not include AI news that users do not necessarily know is AI news, versus just being "on Google". The researchers grouped "preferred sources" so the figure includes everything from podcasts to dailies, and paints the picture of how varied the news publishing landscape has become. The largest first instinct is still to go towards a newsroom for information, but less than it was.
Read

Become a destination, or fade
In 2025, AI bot traffic grew 187% while human traffic grew just 3%. As branded search is falling faster than direct traffic, publishers aren't just competing with each other anymore, they're also fighting against platforms built on individual voices and habit-forming products that are now taken for granted by younger audiences. The way to solve this isn't more of the same, but building named voices that people trust, creating formats that will bring the readers back, as well as investing in the personalisation and registration infrastructure that can turn a casual visitor into audiences worth selling. Harry Clarkson-Bennett has more details.
Read

Making it up as they go
In between countless rulings and legislative decisions which suggest it does not know a law from a hole in the ground, it's good to see that Google's trusted executives can still find time to opine to parliaments worldwide how existing laws it finds unhelpful should be changed. The adjudicated monopolist is again pressuring politicians to change copyright law so AI companies can use existing content without paying, saving the money for new fines perhaps. Google has creatively assembled a white paper which discusses art, music, poetry, written work, and the wonders of strolling through art galleries - all "content", in Google speak, which their people say it should be allowed to take, perhaps lest their share options droop and delay all those artisan cheesemaking start-ups and blog posts about giving back.
Read

The Dutch bot bookshop 
A coalition of Dutch publishers has launched Bookpact.ai, a platform where AI companies can submit proposals to licence books, giving publishers an option to approve or reject requests title by title, with the author's consent. Training, summarisation, translation, and citations are all treated as separate use cases with separate terms. If you want to participate, you set your own price, and if you don't want to, just say no. Major Dutch publishing houses including Maven, VBK, and Lannoo are already on board. Those AI companies which are transparent with the climate impact get prioritised, which is a nice touch. This is something the publishing sector should watch closely, as its opt-in licensing infrastructure is built from inside the industry and not imposed from the outside.
Read

Reach still matters
While Google search referrals to media sites fell 34% year on year, that doesn't mean reach stopped mattering. The data is on the side of bigger brands, as they get more users, however the strategy of shrinking down to "cultivate a loyal core" isn't supported by the evidence. The problem with reach is that the traffic is becoming a bad proxy, and publishers need to sit down and measure whether they come to mind and are present where people spend time, not just who clicked on their links. INMA has the analysis.
Read

Footprints can't hide
After making a fortune off it, Google has published research on how they are now catching and blocking AI slop on YouTube. (Do they just compare it against their own databases?! - Ed) It works on two levels which feed one into another, and in doing so reveal coordinated operations. This could be one of the reasons why publishers saw a wipeout back in January, and it shines a light on the fact that Google can and will detect and kill scaling AI content across multiple properties. The full research can be found here.
Read

Discover giveth, Discover taketh away
Previously we have mentioned how Google is bundling multiple outlets covering the same story into one single card, one example grouping 14 publishers covering the same event under a two-sentence AI summary and a single citation. SEO leaders from Reach, Amsive, and The Times have laid out what's happening and whether preferred sources and publisher profiles are a step in the right direction. They all agree: original content and brand identity matter more than ever, video is becoming unavoidable, and Discover should be considered bonus traffic and not something to build an entire strategy on. Press Gazette has the full brief.
Read

Traffic as a KPI in the year 2026
Valnet has decided to tie their freelancer author pay to the traffic that comes from their articles. Yes, in 2026, this is the way they decided to go. SEO whiz Barry Adams shares his thoughts, and doesn't mince his words.
Read

From inside Glide

Glide Live: the breakdown
The first Glide Live brought together 40+ media and sports brands in a closed room to chat about what actually works in today's publishing landscape. The mood wasn't doom but opportunity: the industry still has trust, relationships, and decades of audience-building that no model can match. Sessions covered SEO, AI, new revenue streams, sports clubs as publishers, and more. Check out the full session summaries for more info.
Read

Big Tech

Addiction back on trial
Both TikTok and YouTube have settled in consolidated social media addiction litigation, now leaving Meta and Snap to face a jury in July. The first trial, from March, ended with a $6 million verdict against Meta and YouTube for negligent platform design - to which they both appealed - and is just one of the many cases which are sitting behind trials, all targeting the same addictive features such as infinite scroll, autoplay, as well as beauty filters that can cause depression and anxiety, among other things. Now we wait for a second jury, and their verdict.
Read

The CMA vs the duopoly 
The Competition and Markets Authority is taking aim at the main app stores' 30% transaction commissions, consulting on users being able to more easily work outside Google and Apple stores when making purchases. The CMA calls the pair a duopoly due to their control of over 90% of UK mobile devices and hurt both consumers and developers. Both Apple and Google could still charge steering fees, but they would have to be justified and fair. This would be an easy win for publishers who have subscription apps.
Read 

AI & Copyright

Hallucinations hurt
A videoconferencing startup is suing Palo Alto Networks after an AI-managed threat intelligence "report" falsely linked the video service to a Chinese hacking operation. Despite the report being de-hallucinated the damage continues thanks to hundreds of security products which share whitelists blocking the company's domains. The entire situation picks at the question of who is liable when automated research leads to lies and consequences.
Read

Ford's 3-point U-turn
Ford had to rehire 300+ veteran quality inspectors after AI-powered checks left literal gaps and holes. The company admitted the bungle and was remarkably non-corpo speak in talking about it, planning to bring back experienced people to improve standards, mentor younger staff and ...err... better train the AI.
Read

People over prompts
In a recent announcement, music streamer Tidal said it won't pay royalties to AI-generated music. AI tracks will not be completely removed but they will be labelled, while royalties are reserved for "original works directly produced, written, and performed by people". Whether they will be able to actually detect and flag the AI stuff is a different question, highlighted by an AI band which released an unflagged Dolly Parton AI cover the day before the announcement.
Read