EU's AI data giveaway, ads in your emails, and media AI Risk Report

Less room for blue links, Meta steps back (for now), and Google now worries about publisher traffic - all in this week's Content Aware.

by Emina Zulic

Published: 15:00, 29 October 2026

Corbidge comments on...dispatches from the content war
From Meta's attempt to snatch everything its business account holders own, to Google's new terms letting AI data from Gmail and Photos feed its ads, our resident war correspondent Rob sees the same front line everywhere: Big Tech blurring the lines of what belongs to us, and what belongs to them. As senators on both sides of the aisle line up against Meta, Rob argues that Zuckerberg seemingly arrived a bit late to the looting party, and to make things even worse, got caught trying to wheel away a stolen fridge down the road.
Read​​​

AI & Copyright

EU's data "gift" to AI giants
EU Member States appear to have fallen asleep while reading their paperwork, or be suffering some malady, after the Council seemingly backed proposals to amend GDPR rules to allow AI companies to use EU citizens' personal data to train models without consent. That would be all data they have, backward-facing, including things sent in messages about you, not by you. Say what? The apparent bloc-wide sop to AI companies is excused on the grounds that any company using AI has an automatic "legitimate interest" in your data in the same manner that, say, a company using your data to deliver your parcels has a legitimate interest. Privacy campaigner Max Schrems of noyb has argued that this is "the biggest gift that EU Member States have ever made to US Big Tech", and promises to take it to the EU's top court if it is approved by legislators. 
Read

Gmail, Gemini, and your ads
While Google's updated privacy policy still says it won't personalise ads based on your Gmail, Photos, or Drive content, it also has a new addition which says "unless you choose to make data from these services available to other Google services that support personalized ads". Personal Intelligence is the exception, but Google does add that the summaries and inferences it draws from them can be used to show ads. This is the first time since 2017 that Google has allowed inbox content to play a part indirectly, and a reminder that "opt-in" AI features rarely stay clean of ads for too long.
Read

Invoiced hallucinations
Western Australia is seeking a refund from Deloitte after its A$166,000 report on merging three of the state's universities turned out to contain AI-hallucinated references, a university name that hasn't been used since 2010, and sums that were off by hundreds of millions of dollars. Deloitte admitted that it used generative AI for research and proofreading, but has dismissed the errors as minor. To make things worse, this isn't the first time this has happened. The firm has already partly refunded Australia's federal government last year over another report that was AI-flawed. Double whammy.
Read

AI regulation, set to music
Since Claude Opus 5.5 came out in September, a new genre of music - made by AI - named "slopcore" has gone viral, with its own fan-run charts, an open-source AI pop star called Claudia, and a surprising number of songs that talk about AI safety, regulation, and the future of work. Copyright and AI law expert Andres Guadamuz, who admits to never being a fan of AI music, has found himself hypnotised by these videos, and he even went as far as to make a guide to those that best explain AI governance, along with a few awkward questions about who the author actually is.
Read

Publishing & Media

The publisher AI risk report
Publishing strategy guru David Buttle's latest AI risk report on how AI Overviews and agents are affecting consumer publisher models is available, and required reading. It looks at just how AI hits different types, some terribly, some actually quite well, and gives a good steer on questions over where to spend your money and effort, and what things are under most threat. He summarises it here for Press Gazette and you can download the full report at the link below.
Get the report

Meta backs down (kind of)
Meta has promised a "full consultation" with news publishers before its controversial new terms come into effect, even if that means they've have to delay them beyond 20 November. This decision came after ten Norwegian media groups won a temporary court order blocking the terms until their legality has been checked, and amid letters to Zuckerberg from both Democratic and Republican senators. While this is a welcome sight, it's still not a contract, and until Meta puts it into writing that it won't scrape publishers' websites, the fight is still ongoing.
Read

Show them what's popular
With clicks getting harder to come by, SEO guru Harry Clarkson-Bennett points to a simple trick from The Athletic's homepage: showing the number of comments that selected articles have generated. If the comment count is high, then there is a conversation worth joining, and that is a reason to click that goes beyond headlines and images. Start with comments, and then experiment with shares, likes, and other engagement signals to spot the articles that your audience values, and also give them a bigger spotlight on your homepage. 
Read

Three questions for your CMS vendor
Many CMS vendors trumpet that AI is "built in, not bolted on" - but how do you know? Glide CTO Nedim Dedic sat down with tech title Ingedelo to explain ways to tell the difference and suggests three scenarios to put to any vendor: does the AI make changes the same way an editor does, does it understand your content model or just the text it sees, and can it change anything without an editor's approval? He also shares what's surprised the team most, and why flashy features lose out to the small tools which take care of editors' mechanical work in the newsroom popularity stakes.
Read

Live and kicking
With referral traffic sinking, live blogs give publishers a reason for audiences to come straight to them and also keep coming back. It doesn't matter if you're adding value to subscribers like The Times, or you're building reading habits and pointing audiences to your reporting like The Guardian, both of these tactics turn live updates into longer-term relationships. 
Read

Setting the record straight
As there is more misinformation than ever, setting the record straight has become a regular part of local reporting, from tracking down deceptively edited videos to correcting viral myths. Newsrooms are explaining why they're investigating a story, answering misinformation in their own comments, and inviting readers to send in tips directly. It's a good reminder in Media Literacy Week that trust is earned one fact at a time.
Read

Certified human
With AI slop flooding online bookshops, publishers overwhelmed with AI submissions, and literary prize winners accused of using AI, "human-authored" certification schemes are slowly appearing around the world. One certification company believes it could become an industry worth billions, but it's the authors and publishers who will be footing the bill, not the tech companies that created the problem. Meanwhile, a prize judge warns that AI detection tools are being used to make accusations that could tarnish a writer's career, even when the investigation shows that the work was human after all. 
Read

Overviews, everywhere
Google's AI Overviews now appear on more than 80% of searches according to some trackers, which puts them up from around half in June last year and about 15% the year before. Google is showing them for brand searches and even for fake or incorrect queries, as part of Google’s push to have an answer for every question. Google's recent link changes suggest it knows AI search still needs the open web, but with Overviews this widespread, the space left for a plain blue link keeps shrinking.
Read

From inside Glide

From anonymous to account
By the time someone has registered on your site or app, they may have already visited regularly, read countless articles, voted in a poll, and checked the subscription page multiple times. Glide Nexa can recognise anonymous visitors and compile a pattern of what they like - and connect it to their profile when they register and agree to your terms. That gives teams a clear view of which content and touchpoints actually drive sign-ups, and how long the journey takes.
Read

One view of the whole desk
At the start of their shift, a desk editor needs to know what's in draft, what's waiting for review, and what's scheduled to go. Content Queues in Glide CMS puts all of that on one dashboard, with personal queues for an editor's own work and shared global queues that give the whole team the same picture. Each editor arranges their dashboard the way they like, and queues refresh as work moves, so nobody has to ask what's changed.
Read

Big Tech

US state goes after Meta
Rhode Island is suing Meta, alleging that it knowingly profited from scam ads on Facebook, Instagram, and WhatsApp, and that its own AI advertising tools helped scammers make them. The complaint is leaning on internal documents that show Meta estimated it served users around 15 billion high-risk scam ads a day and expected about 10% of its 2024 revenue, which is roughly $16 billion, to come from scams and banned goods. Meta claims it is fighting fraud with all it's got, but after New Mexico's privacy case, it's another state that thinks courts can get it further than Meta's promises. 
Read

Google cries foul
According to Google, the UK competition watchdog's plan is "unlawful" because the CMA proposed letting AI assistants such as ChatGPT appear on a yearly choice screen on Android phones and Chrome. It argues that AI assistants were never part of the original investigation, and it also warns that steering people towards AI would hurt publishers (suddenly Google cares about publishers), since AI sends fewer visitors to news sites than traditional search. It would've sounded more convincing if it weren't coming from a company that runs AI Overviews on most of its searches. 
Read