The great traffic vanishing act, no AI rights for free, and it's all kicking off on Copilot

Discover-ing scams, Australia cracks down on social media, Google's data doctrine, and comments making a comeback - all in this week's Content Aware.

Published: 15:03, 15 January 2026

Corbidge comments on...AI policies, brought to you by lobbyists
Public policy debates around AI continue to resemble corporate press conferences where lobbyists muscle out journalists. As the prospective legislation train rumbles forward, and the UK government is reconsidering its preferred opt-out approach to AI training on copyrighted work, our critic in chief Rob has been tuning into the latest committee meetings on who should pay what for copyrighted work.
Read

Trust over tech redub
Newsrooms experimenting with AI got a clear memo from their audiences: humans must stay in the driver's seat. Two new surveys reiterate previous findings that the public think AI in news in a bad thing. In the US a survey of around 1,500 local news consumers found that while people aren't anti-AI, they want transparency, accuracy, and ethical guardrails wherever it is used, and that applies especially to creating content. In the UK, 4000 consumers were surveyed by Newscast and highlighted plummeting trust in AI-generated content, but also a rise in trust in the role that real journalism plays. No one has any issues with AI doing the grunt work when it is a trusty sidekick and not a suspicious interloper.
Read

Google turns off the tap
According to Chartbeat data, search referrals to publishers are down by roughly a third worldwide, with even Google Discover taking a hit. The culprit? AI Summaries, hogging the top results. In the US, the drop is even steeper, with lifestyle staples such as weather and TV guides feeling the hit first. Publishers aren't optimistic things will improve and expect traffic to fall another 40+% over the next few years, leading them to plan on easing off Google, Facebook, and X as a result and focus on subscriptions, content-creator vibes, and influencers. Press Gazette shares more details.
Read

Journalist to algorist
Traditional media presentation styles are being refashioned by consumption habits, as influencers win trust with more personable styles and younger audiences are led by social platform algorithms. Reuters Institute research shows established newsrooms are leaning into more presenter-led and distinctive journalism, pushing harder into video, making flexible content, and cautiously using AI to work faster and smarter. It all might sound a bit frothy to those who prefer to stay out of the limelight but standing in the dark has its downsides. Download the full digital news project here [PDF].
Read

Comments strike back
Titles like The Washington Post, the Financial Times, Wired, and The Times are all reopening comment sections, moderated and mostly for paying subscribers. Having previously stepped away the effort of moderation managing the occasional chaos of comment, many sites are showing that when done right it's a loyalty machine that keeps readers engaged, subscribed, and invested.
Read

The AI game is no game
The UK's West Midlands Police force found out the hard way that AI doesn't differentiate between football fact and fiction. A controversial intelligence report, used to justify a travel ban on fans from Maccabi Tel Aviv FC to a Europa League match in the force's territory, was found to have quoted crowd incidents from entirely fictional previous matches courtesy of Microsoft Copilot - leading to calls for the force's police chief to quit or be sacked. Microsoft does warn that their AI "can make mistakes" but this wasn't just a typo, it's real-world security decisions by police based on AI hallucinations.
Read

It's OUR content now
Google has made its view of the use of content for training AI crystal clear to UK lawmakers: web content shouldn't have to be paid for. According to UK Googler (and ex-Ofcom) Roxanne Carter, Google's AI models aren't copying publishers' work but simply crunching massive datasets to spit out "wholly new content". She explained that the company is doing deals for archived materials, specialised datasets, or off-platform access, but wobbled on clarifying how opting out of AI scraping might hurt Google search results in Discover or Overviews.
Read

Click at your own risk
Google Discover's cosplay as a virus delivery services continues, despite claiming it would tackle spam site recommendations. Users continue to report regular pop-up warnings on devices after clicking sites promoted by the feed, claiming the platform is quietly steering users towards risky corners of the web.
Read

One editor to rule them all
Publishers have long wrestled with clunky workflows, platform switching, manual embeds, and wrestling metadata, just to be able to get a video on their page. Enter Glide CMS and EX.CO, riding in together to stich video management right into the editor. Now teams can easily embed, optimise, recommend, and even monetise video without needing to leave an article. Metadata got smarter, recommendations get sharper, and revenue looks healthier. No more video chaos!
Read

Write for readers, not robots
Google said it loud and clear: don't slice your content into bite-sized chunks just to impress AI or LLMs. While it might give a temporary boost, the systems are evolving and learning to reward content written for humans and not for machines. As the algorithms change and evolve, writing content for AI could backfire, and in the long run, you should aim to impress your readers and not a robot in a data centre.
Read

Prompting ≠ authorship
Warner Music has teamed up with Suno to let its catalogue feed their AI-generated songs - but if you use it to make a smash hit, make sure to read the fine print. Basic users now don't own the music they create as Suno only grants a license and Warner/Suno keep the rights. Paid users can do what they want, which sounds remarkably like a short-lived xAI Grok ditty. Opinions on AI music aside, the deal may give some indicator to what future content deals with AI companies may look like in the real world.
Read

Meta cleans house down under
In order to comply with Australia's under-16 social media ban, the company has axed nearly 550,000 accounts across Instagram, Facebook, and Threads. Besides that, activity-based age checks, selfies, and other verification checks are now part of the daily moderation mix, all to avoid fines which could go up to AUD 49.5 million.
Read

Gemini gets personal
Gemini seems to be turning into some sort of stalker, albeit a polite one. The new Personal Intelligence beta lets the assistant tap into your Gmail, YouTube habits, photos, search, and more, all in hopes to deliver more hyper-personalised answers. You have the option to control what it sees (at least something), but things may get awkward, inaccurate, or too personal. For now, only paying US users have it, but a wider release is expected.
Read