A myth terrorising publishers, search is wider than you think, and the comeback of real news

The guardrails are coming off, AI isn't reading your instructions, Google's 90-day window, and Microsoft's Discord goes dark - all in this week's Content Aware.

Published: 15:31, 05 March 2026

Corbidge comments on...the danger of content sausage factories
What makes a news organisation worth paying for in the first place? Of course it's about the facts - but if it is only about facts, then the risk is that your output isn't differentiated from anyone else who can appear to put a real or virtual boot on the ground as a story breaks. As major news orgs face the challenge of what their worth is to real people and also to AI, our glitch in the matrix Rob weighs the conundrum of human voices in a robot-ready world.
Read

Publishing & Media

The Google Zero mirage
For all the talk in media that Google Zero is the doomsday prophecy for traffic coming to sites from Search and Discover, not everyone agrees that it is even a thing. Global SEO voice Barry Adams cautions that while there is a rapid spread of belief in a cool-sounding term, it doesn't stand up to scrutiny. Accepting it as inevitable is severely damaging to the industry, and risks a doom-spiral where brands turn down the heat on SEO at the very moment it is becoming of escalating importance. If Google Zero is the fire in the distance, you don't reduce the supply of water to the fire hoses - especially if it is a mirage not a maelstrom.
Read

A search by any other name
When is a search not a search? Perhaps when it happens anywhere other than Google. A new study of 41 websites suggests the volume of searches taking place outside of the usual "google it" space is far larger than thought. Just because a search happens somewhere like Amazon, YouTube, Instagram, Reddit, and dozens of other platforms, doesn't mean it's not a search. When other places are factored in, Google's share of search drops from 90% to 70% in the US, and "big" new rivals like ChatGPT drop below Amazon, Bing, and YouTube. The takeaway is pretty straightforward, don't just optimise for Google and ChatGPT, optimise for where your audience is looking.
Read

The reporting revival
While it sometimes seems as if journalism is walking a tightrope over the AI pit, there is no need to cue the dramatic music. Readers are still showing up and opening wallets, signalling that humans still matter and that they prefer actual, human reporting. With audiences already tired of AI slop and clickbait, original and trusted journalism is making a comeback and with mission-driven owners, distinct newsroom voices, and some industry teamwork, the glass that was once half-empty now feels a bit more like it's half-full. Quality reporting is getting back into the spotlight, with humans firmly in the drivers seat.
Read

No human, no copyright
The US Supreme Court has seemingly settled the question of whether AI-generated art can be copyrighted by declining to even hear a case arguing that it can. It reasserts that human authorship is a requisite for copyright protection. Both the US and the UK Supreme Courts have agreed on the conclusion, as while AI can be a tool in the creative process, a machine cannot own what it makes.
Read

Brace for more ranking turbulence 
Since the start of the year Google rankings have been in constant flux and subject to noticeable volatility. We were gifted a major core update which wrapped up at the end of February, but the turbulence has yet to settle: smaller, unannounced updates appear to be rolling out on a near-daily basis. The ground isn't likely to stop shifting anytime soon, so keep in mind that what worked last week might not work next week. Search Engine Roundtable has more details.
Read

Keywords are so last week
According to new research (PDF download), Google's novel STATIC framework (Sparse Transition Matrix-Accelerated Trie Index) allows its AI-powered search to be a 1,000 times faster than existing widespread methods. By applying filters such as freshness, location, and content quality, and applying what they call semantic identity, the approach allows you to signal to search more information, quicker, for less cost. While it isn't confirmed live yet, according to Google's Labs the pieces are already moving.
Read

Publish once, reach all
A breaking story shouldn't take longer to distribute than it did to write, but for a lot of editorial teams this is exactly the case: production trumps creation. Copying, pasting, and reformatting the same content across websites, apps, newsletters, and social feeds eats up hours that should go into finding and creating the things people want most. Glide CMS makes sure there is none of that, as the multichannel publishing feature lets editors write once and push everyone from one workspace, while automation handles routine distribution and editorial control stays intact where it matters most.
Read

Big Tech

The 90-day clock waits for no one
Google's three-strikes system can chop down your advertising account even before you even realised you had done something wrong. Three policy violations within 90 days, and you can be out - however, as one user discovered, you don't even need to actually break any rules to get hit. After being hit with strikes and having two appeals rejected, one site managed to get the ban reversed after plastering their site with disclaimers. The lesson? Know Google's policies, remove ads you're not using, and always over-explain your situation at every opportunity.
Read

Streisand effect, Microsoft edition
Microsoft's relentless AI push might be winning friends at hedge funds and data centre energy firms, but seemingly less so amongst users. After consumers dubbed questionable AI features "Microslop", and Microsoft tried to wipe its presence from user-facing groups and channels, the Streisand effect did the rest.
Read

AI & Copyright

404: content idea found
AI hallucinations are all bad, right? One SEO team accidentally turned them into a content strategy. After looking at their blog for broken pages, the folks at Ahrefs found that URLs hallucinated by AI tools actually revealed gaps in their publishing strategy. Now articles are being written to match these hallucinations, attempting to turn dead ends into real backlinks.
Read

The file that changes nothing
LLMs.txt files supposedly offer a way for websites to shape how AI models perceive and reference their sites, though plenty disagree that they serve their purpose well because they can be ignored. The team at Reboot found that AI bots aren't even checking for the LLM.txt files, and little evidence they actually act on it when they do find one. What actually works is sticking to the basics: strong internal linking, well-optimised content on pages AI already visits, and getting your brand mentioned on third-party sites that AI models are already citing.
Read

Wikipedia vs foreign lies
After discovering that use of AI to translate articles caused hallucinations to appear in translations, such as books being cited for things they never said, and paragraphs sourced from unrelated materials, Wikipedia editors threw down the restrictions hammer. The organisation's response? Put translators on a short leash, and - for now - get other AIs to check the work. To tediously insert a shared lesson from the GPP CMS crew, it's why we do not use direct AI translations in our CMS, and revert to less glamorous but less fictionally-capable Machine Learning methods instead, because it's harder to spot hallucinations when they are in another language.
Read