UK watchdog splits the Google vice, NYT's Marseille warning, and why "have your say" matters again

AI triage saves lives in the UK, Altman faces a jury trial in Florida, and Meta builds a template for automating your job - all in this week's Content Aware.

Published: 14:59, 04 June 2026

Denis demands... action on AI before the free rides stop
With regular Content Aware opinionmeister Rob Corbidge monitoring matters in the US, Glide's alternate voice of reason Denis Haman eyes the dangers of subsidised addiction to AI, and the pincer of threat to publishers and other users who have a short window of time to act before opportunities to claw back money from AI bandits disappears.
Read

Publishing & Media

Tsunami warning from Marseille
New York Times publisher AG Sulzberger used his WAN-IFRA keynote in Marseille to urge publishers to stop being "too quiet, too passive, and too fragmented" while AI companies steal their work. The NYT has spent more than $20m on lawsuits against OpenAI/Microsoft and Perplexity since December 2023, while spending $2bn in 2025 alone producing journalism. US private AI investment hit $350bn in 2025, but less than half of one per cent went back to the people creating the data that powers it. Sulzberger's message: stand up for yourself, deal carefully, push for legislation, and make sure people seek your journalism directly. Press Gazette has more.
Read

UK publishers get a Google opt-out button
The UK's Competition and Markets Authority has ruled that news organisations can now block their content from appearing in Google's AI summaries without losing their place in traditional search results. Until now, publishers had to choose between letting Google scrape their work for AI Overviews or disappearing from 90% of search entirely. Google plans to roll out the new controls globally, which suggests the CMA's intervention carries weight beyond the UK and Google cannot afford to ignore regional regulators. Whether publishers actually use the opt-out is the question everyone will be watching.
Read

Not every update is breaking news
Most CMS platforms treat a typo the same way they would a breaking news story: same notifications, same timestamp resets, same webhook fires. At scale, that means a lot of readers re-alerted, RSS feeds duplicated, and analytics skewed. Glide CMS gives editors three publishing behaviours per update: live timestamps for genuine new content, silent update for routine fixes, and manual timestamps for right-dating content to real-world events. The choice is yours and no developer is involved, while the entire pipeline respects it. One story with five updates stays one story, not six.
Read

Plugin hands attackers the keys
WP Maps Pro, a commercial WordPress plugin with over 15,000 sales, has a critical vulnerability being actively exploited to create rogue admin accounts on vulnerable sites. The flaw scored 9.8 out of 10: any unauthenticated visitor can call a "temporary access" endpoint, create an administrator account, and take full control of the site without credentials or social engineering. Wordfence blocked nearly 3,000 attacks in a single day. As the plugin isn't sold through a WordPress directory, updates don't arrive automatically and many site owners won't know about the patch unless they're looking for it. Anyone with WP Maps Pro should update to version 6.1.1 immediately, or disable the plugin.
Read

Regular updates, irregular chaos
Google's May 2026 broad core update has finished rolling out, with significant ranking volatility hitting across multiple weekends and a final spike on completion day. The update felt longer than the March core update, and Google described it as a "regular update designed to better surface relevant, satisfying content for searchers from all types of sites". How much of that is true remains to be seen.
Read

WoodWing Assets meet Glide CMS
Most enterprise DAMs sit outside the CMS, which means editors need to download images daily, lose metadata in transit, manually re-upload everything to the CMS, and watch rights information disappear along the way. Even when the DAM is meant to organise assets, the disconnect between systems turns it into another workflow bottleneck. Glide CMS connects directly to WoodWing Assets, letting editors search and import from inside the CMS with metadata intact. WoodWing handles rights and governance, while Glide CMS handles publishing variants and optimisation, and the gap between them stops being a problem. For publishers already committed to WoodWing, this removes one of the most tedious parts of daily editorial work.
Read

Comments are the new product
With declining traffic and AI on the rise, several major publishers including the New York Times, Financial Times, and Der Spiegel are turning their comment sections into genuine editorial products. Reporters are joining discussions, AI prompts questions, branded debate spaces emerge. In a world of fly-by readers, the ones who interact and talk and advocate are the business model.
Read

Google fail = AI fail
The worst thing you can do for your AI search visibility is wreck your SEO. Sites losing Google visibility start dropping in ChatGPT citations soon after. However, many recommendations for GEO are things SEO veterans would flag as future liabilities - shortcuts that work until search engines treat them as spam. Google employees are already publicly discouraging things like "inauthentic mentions" and LLMs are taking steps to clean up citation manipulation. The best approach is the one that leaves your brand cited after the spam cleanups are done. Lily Ray shares more.
Read

Big Tech

Template for replacing human work
As we've mentioned previously, Meta has been on an AI workplace monitoring push, and now internal documents reveal the full scope. It is not only looking at mouse movements, keystrokes, and screens, but also logging how employees use their work computers across more than 200 apps and websites, all to train AI agents that can carry out routine tasks autonomously. Employees have reported that the tool consumes an entire month's home internet allowance in days, describing Meta's goal as building an AI that knows which dropdown to click, which document to paste where, and what to do next. Privacy experts warn that this thing is bigger than Meta, as it's a template for converting any human workflow into a machine-readable system. Might not be a bad idea to have Sarah Connor on speed dial.
Read

Florida aims at OpenAI
Florida's attorney general has filed a lawsuit against Open AI and Sam Altman personally, alleging OpenAI marketed ChatGPT as safe while knowing that it can cause serious harm. The lawsuit argues that the product was designed to be sycophantic, agreeing with users roughly ten times as often as disagreeing, in order to keep them engaged. According to internal documents, there have been warnings as early as 2023 that the model could coach people on committing crimes, but Altman decided to ignore them. The lawsuit names him personally, holding him liable for "utter disregard for the risk to human life". The state is demanding a jury trial, and the penalties could go into billions.
Read

Microsoft targets 'make people addicted' to AI
Internal documents for Microsoft's new AI assistant, Scout, reveal the company's strategy in plain language with "Make people addicted" spelt out. The plan has three phases for the agent, starting with addiction, then connecting to more AI services, and then adding more features. The leak came at a perfect time, as there is a growing scrutiny over AI chatbot dependency, with recent research showing they can fuel delusions and dangerous behaviours.
Read

Google ad brought nasty surprise
Malicious Google ads are pushing FlutterShell backdoor through fake Mac productivity apps that sailed through Apple's security screening. The malware hijacks Chrome, rerouting traffic through ad-filled proxies while handing attackers shell access. The apps carried valid Apple Developer IDs and were distributed via Google-verified shell companies.
Read

AI & Copyright

Same manipulation, new wrapper 
Researchers at the Center for Democracy & Technology studied how chatbots from OpenAI, Google, Meta, Replica, and Character.AI use manipulative tactics to keep users engaged, such as sharing data, and spending money. They identified 37 dark patterns, from bots pledging secrecy to companion apps guilt-tripping users. And for the old tricks, they're still here, just evolved: instead of infinite-scroll, now there is a follow-up prompt after every response; and chatbots which mirror your values back at you. The format has changed, the incentive hasn't.
Read

One smart duck
Since Google announced its AI-first search overhaul in early May, people already started to move elsewhere. One of their destinations is DuckDuckGo, who has seen app installs jump 18% and visits to its AI-free search page rise 23% week-over-week. The privacy-focused engine is now leaning hard into being the refuge for users tired of seeing their data used everywhere and AI in everything.
Read

When agentic AI gets it right
43% of UK voters don't trust the government to deploy AI responsibly, but Thames Valley Police's Bobbi assistant shows what happens when you get it right. The AI triages non-emergency digital contact and has helped identify urgent domestic abuse and rape reports when victims couldn't safely call 999, escalating them to human operators who dispatched officers. The force tested it with victim support groups before deployment and changed the system based on their feedback.
Read