Game Over
If there’s one thing Google seems to excel at, it’s operating a monopoly. First it was search, then ad tech, and now AI is the next obvious frontier.
But, this time, regulators are trying to get ahead of the game.
On Thursday, the EU ordered Google to scrap restrictions on how rival AI services operate on Android. Under the ruling, Google must open 11 Android features to competitors so they can “access key functionalities and better compete with Google’s Gemini AI service,” Reuters reports, including activating AI assistants with a voice command and delegating specific actions within apps.
The order also mandates that Google share the data it uses to optimize its search services.
Google, unsurprisingly, isn’t thrilled.
The decision could “[undermine] vital privacy and security guardrails for millions of Europeans,” Kent Walker, Google’s general counsel, said in a statement. But, per Reuters, the European Commission claims that the measures “contain robust safeguards to protect users’ privacy and device security” and that Google’s features will only be offered to “rivals who fulfill security and privacy criteria.”
Meanwhile, the fights over AI control keep piling up. Last week, Google sued OpenAI for purportedly stealing company secrets.
Well, soon the secrets might not need to be stolen – they’ll be up for grabs.
Sorry, I Can’t Do That, Dave
A new report by the Meta Oversight Board suggests that large language models like Claude and ChatGPT are more likely to resist generating content critical of authoritarian governments, Fortune reports.
In other words, if a user asks an LLM to generate protest materials against a typically permissive country, like the United States, it will do so. But if they request protest materials for a more restrictive country, like Saudi Arabia, it will refuse.
Although the cause is unclear, researchers suggest that the models are picking up on a lack of available training data related to repressive regimes. It’s also possible that the requests are coming up against AI guardrails, as many LLMs would cite legal or safety reasons for not complying.
Obviously, most advertisers would not be using AI for such explicitly political purposes. But considering how many products are built on LLMs, it’s worth examining how they reflect existing power structures in unanticipated ways.
For example, would a chatbot be more likely to recommend a larger brand to a user over a smaller one based solely on how often the former appears online? And if a brand routinely issues takedown notices to its critics, will that lack of public criticism be reflected in AI searches?
Don’t Hate, Generate
A great deal has been made lately of digital media and traditional news publishers giving Google the cold shoulder by blocking its web scraper for search and AI training data.
And, hey, why not, considering the traffic spigot is already dwindling to a trickle?
But let’s be real. It’s not like any news company can really quit Google – or opt out of LLMs and AI search engines, which is where Google is headed anyway. Instead, they’re trying to find ways to make this new reality work for them.
Digiday reports on a growing cohort of publishers, including Axios, Forbes, Future, Time and The Washington Post, that are productizing generative AI responses as part of commercial deals with the brands or services they write about. Think content marketing listicles or other assorted sponcon.
This idea is a natural next step – or perhaps a desperate fallback – for companies with affiliate marketing businesses, like product review sites, that are being squeezed by these same AI search trends.
“Publishers aren’t just selling impressions anymore,” says Forbes Chief Innovation Officer Nina Gould. “They’re selling visibility within the AI knowledge ecosystem.”
But Wait! There’s More!
Shopping startup Phia is being accused of outright affiliate hijacking, not garden‑variety fraud – and marketers should care about the difference. [Adweek]
Internet users have started to get fed up with all the “opt-out” defaults for AI features. [Wired]
Beehiiv is apparently trying to become a social network as it now lets newsletter subscribers chat with each other. [TechCrunch]
Over the past two years, YouTube has gotten slightly better at not serving harmful eating disorder content to teenage girls, according to The Center For Countering Digital Hate. [Tubefilter]
On a related note, parents who use Instagram supervision controls will now be notified when their children start talking to Meta AI about suicide. [Mashable]
Third‑party app stores are coming to Google Play next week, as the years‑long Epic Games v. Google antitrust case finally nears its end. [Ars Technica]
An ideological fight over how to regulate AI – with rank‑and‑file employees in open conflict with company leadership – is driving up political ad spend, as each side backs rival super PACs. [Wired]
You’re Hired!
Vinny Rinaldi, a Hershey’s marketing vet, joins The Trade Desk as VP of client strategy & growth. [release]
Here’s today’s AdExchanger.com news round-up… Want it by email? Sign up here.
