When Google not too long ago claimed that its AI Overviews are driving “extra queries and better high quality clicks”, it struck a nerve. Too many, it wasn’t simply out of contact; it was patronizing, one other reminder of how far the platform’s incentives have drifted from theirs.
The message got here through a weblog submit from Liz Reid, Google’s head of search, who painted an image of stability: natural visitors holding regular, click on high quality enhancing and publishers supposedly getting extra worth from search than they did a yr in the past. As for the rising pile of third-party reviews exhibiting steep drops in visitors? Reid waved them off as flawed — constructed on dangerous knowledge, edge instances or traits that began earlier than AI Overviews even launched.
However Google isn’t alone in rewriting the story. Throughout platforms, the identical playbook is in use: soften the language, reframe metrics, present as much as the odd trade occasion with platitudes. This information unpacks the tells — learn how to spot the spin earlier than it turns into the story.
“Belief us, visitors is definitely higher now”
Get used to listening to that. Google’s newest weblog submit was much less a protection and extra a smooth spin: AI Overviews aren’t tanking visitors a lot as they’re enhancing it.
Certain, there’s a case to be made: a click on after an AI abstract is likely to be extra significant since they’re prone to be extra engaged. However for publishers watching referral visitors crater, it’s a hole argument.
The takeaway: when platforms begin redefining what success seems like, it’s normally as a result of the previous metrics look ugly. Belief your individual knowledge, not their narrative.
“Affordable licensing phrases” means take it or go away it
One of the crucial dependable pink flags in AI licensing bulletins isn’t buried in legalese — it’s proper there within the language.
Take this phrase: “affordable licensing phrases”. It reveals up typically, and it sounds…high-quality. However what it normally means is: “we’re paying simply sufficient to name this a deal, not sufficient to match the worth we’re extracting”. The uncomfortable reality is that there is no such thing as a actual market charge for AI use of writer content material — solely the one set by the platforms. Most publishers don’t have the leverage to outline what “affordable” even means.
The takeaway: look ahead to euphemisms. Phrases like “affordable” and “truthful market worth” aren’t impartial — they’re strategic. They reframe energy imbalances as partnerships. Be taught to decode them. Or when you can afford to, rent an professional to take action like The New York Instances, Washington Put up and Gannett have.
But when you’ll take the verify, accomplish that realizing its smaller than it seems
We get it. Typically you are taking the verify. Simply go in clear-eyed: short-term cash not often solves long-term issues in media. These offers would possibly easy issues over for a minute however they don’t essentially repair the underlying cracks. If something, they normally go away extra questions than solutions.
As one one industrial exec put it on the situation of anonymity: “Are these offers supposed to be a gesture to the information neighborhood, which goes to worsen lots of of websites and media corporations and fulfill 20?”
Which is why some publishers are taking part in it sluggish, ready to see whether or not an actual market round these offers emerges. Optimists suppose platforms will come round. Cynics are betting on lawsuits or regulation to power their hand.
The takeaway: Solely a handful of publishers will get actual cash — Information Corp is getting $50 million a yr from OpenAI, for example. The remaining will doubtless get gestures. And whereas these checks would possibly assist now, they received’t rebuild what AI is eroding.
“Your Content material Is Powering the Future (However you’re not invited to form it)”
Publishers have lengthy feared that their content material would change into uncooked materials for another person’s roadmap. The platform has carried out little to show them incorrect. Each few years, a brand new product cycle rolls round, the sample repeats: media fuels the machine however by no means will get to steer it. The AI growth is the most recent iteration of that dysfunctional flywheel.
Living proof: OpenAI’s Nick Turley not too long ago stated that the corporate’s partnerships are supposed to “form the way forward for content material”. Days earlier than, OpenAI skipped an IAB Tech Lab occasion the place 80 media execs gathered to debate precisely that. The optics aren’t delicate.
The takeaway: Platforms can pay publishers for content material, however as for giving them management? That stays of their palms.
Exhibiting up shouldn’t be the identical as exhibiting transparency
Platforms are getting higher at optics. They’ll do the trade occasions, they’ll be part of the panels, they could even sponsor the convention. However don’t confuse presence for openness.
Have a look at Meta eventually month’s IAB Tech Lab occasion final month. The exec who introduced there – Shelley Venus, Meta’s senior director of worldwide media technique – was polished, on-message and stated all the appropriate issues for the room, based on two execs who have been there. However, as at all times, the proof isn’t within the pudding.
“[Venus] was clear that the brand new management is aware of that AI runs on good content material, and that has moved Meta to have interaction extra [with publishers],” a kind of execs informed Digiday, underneath the situation of anonymity. “TBD in fact on what that appears like.”
“We stay dedicated to supporting high quality journalism”.
That is the closing line of numerous weblog posts, press releases and on-stage remarks. However like several good PR line, it’s so imprecise it’s functionally meaningless.
Supporting journalism can imply a licensing cope with a handful of publishers, or a PR grant to fund newsroom innovation. However it could actually additionally coexist with algorithm shifts that intestine referral visitors and AI merchandise that summarize articles with out sending customers to them. Dedication, on this context, doesn’t at all times include accountability.
The takeaway: Have a look at the incentives, not the statements. If a platform’s enterprise mannequin thrives whereas yours erodes their “help” isn’t a method — its a slogan.
