Google Just Learned France Doesn't Negotiate AI Overviews for Free

You've been hearing "AI Overviews are killing publisher traffic" as a complaint for two years now. It usually stays a complaint. On August 11, it became a legal filing, and the group filing it picked a country with a specific, on-the-books reason Google can't just wait out: a 2022 commitment Google itself made about paying for exactly this kind of content use.

What actually happened on August 11

APIG, the alliance representing nearly 300 French newspapers, filed a complaint with France's competition authority, the Autorité de la concurrence, alleging that Google deployed AI Overviews and AI Mode in France on July 22, 2026 without the prior consultation or payment negotiation its own earlier commitments require. This isn't a press release dressed up as news. It's a formal regulatory filing, the same mechanism France already used successfully against Meta a month earlier.

The gap between launch and complaint is worth sitting with too: twenty days. Google turned the feature on in France on July 22, and by August 11 a coalition representing nearly 300 outlets had already coordinated a legal filing citing a specific, dated breach. That's not the pace of a slow-building grassroots grievance. It's the pace of a group that had the legal argument prepared before the feature even launched, watching for the trigger event rather than reacting to years of accumulated frustration. Publishers have had two years of AI Overviews rollouts elsewhere to learn what the traffic impact looks like, and by the time it hit their own market, they were ready to move immediately rather than negotiate quietly first.

The 2022 commitment Google allegedly broke

The legal hook here matters more than the traffic numbers, because traffic-loss complaints alone haven't moved Google before. APIG's argument is that Google made specific commitments back in 2022, under France's neighboring-rights framework (itself derived from the EU Copyright Directive adopted in 2019), that included transparency obligations and remuneration tied specifically to AI services. Deploying AI Overviews and AI Mode without opening negotiations under that existing framework is, in APIG's reading, not a new dispute. It's Google breaking a promise it already made in writing.

“This is not publishers asking for something entirely new. It's publishers pointing at a commitment Google already signed years ago and asking why AI Overviews didn't trigger it the moment the feature went live.”

That framing changes the burden of proof. A regulator deciding whether a brand-new AI feature deserves brand-new payment rules has a lot of room to stall, study, and negotiate slowly. A regulator deciding whether a company honored a commitment it already made in 2022 has a much narrower, more mechanical question in front of it, and narrower questions tend to move faster through competition authorities than open-ended ones.

It's also a template other markets can copy without having to build a new legal theory from scratch. Any jurisdiction where Google has made a comparable neighboring-rights or content-licensing commitment, ahead of an AI feature that arguably triggers it, now has a working example of how to frame the complaint, who to file it with, and what evidence to bring. Legal arguments that work tend to get reused, and competition authorities in adjacent EU member states read each other's enforcement actions closely. A French win doesn't just resolve a French dispute. It hands every other regulator holding a similar 2022-era commitment a tested playbook.

Why this isn't France's first swing, or its last

France's Autorité de la concurrence isn't a first-time actor here. On July 8, 2026, a month before the APIG complaint, the same regulator ordered Meta to resume negotiations with publishers over payments related to AI content use. APIG's complaint against Google is explicitly modeled on getting the same kind of order applied to a second company.

Laid out on a timeline like that, the pattern is hard to miss: a formal EU-level investigation already open, a successful enforcement action against a comparable platform a month prior, and then the same regulator asked to apply the same logic to Google within three weeks of the feature actually going live in-market. That's not a coincidence of scheduling. It's a regulator that's found a playbook that works and is running it a second time while the first case is still fresh.

The Meta precedent is the part of this story that deserves more attention than it's gotten. An order to "resume negotiations" sounds procedural, almost toothless, until you notice what it actually implies: a major platform was found to have stopped negotiating with publishers over AI content payments, in a market where a regulator had the standing to force it back to the table. That's a meaningful signal about how these disputes get resolved once a regulator engages seriously. Nobody expected Meta to reach a payment agreement through a strongly worded statement. It took an enforceable order. APIG is betting that the same mechanism, applied to Google, produces the same kind of result.

The traffic numbers behind the complaint

APIG's complaint leans on an estimate from Arcom, France's audiovisual and digital communications regulator, putting the traffic loss attributable to AI-generated summaries at 33% to 38% in European markets where the features are already active. That range sits inside the broader pattern this publication has tracked all year: Penske Media's U.S. antitrust filing cited click-through declines up to 58%, and our own analysis of AI Overviews traffic loss found the same directional pattern across multiple independent measurements, even when the exact percentage moves depending on category and market.

The consistency across independently sourced numbers, French regulatory estimates, U.S. antitrust filings, and Something Inc.'s own tracked data, is what makes this complaint harder to wave off as one aggrieved trade group's exaggeration. Multiple parties, in multiple jurisdictions, using different measurement methods, are landing in the same rough neighborhood: a third or more of the traffic these AI features could otherwise have sent is simply not arriving.

It's worth being specific about what that traffic actually represents, because "33 to 38 percent" can read as an abstraction until you translate it into what a newsroom does with it. That's subscription conversions that never happen because a reader got their answer without visiting. It's display and programmatic ad revenue that scales with pageviews and doesn't scale when the pageview never occurs. And it's the slower, harder-to-measure loss of a reader relationship: someone who used to visit a specific outlet regularly now gets a synthesized answer with no memory of whose reporting it was built from. Traffic loss is the easy number to cite in a regulatory filing. The relationship loss underneath it is the harder, longer-term cost, and it's the one that doesn't reverse even if a payment settlement eventually gets negotiated.

APIG's complaint is also a bet on timing that's easy to miss if you only read the headline. Filing now, twenty days after launch and with the Meta precedent still fresh, puts pressure on the Autorité de la concurrence to act while its own recent order is the most visible and least contested. Wait six months, and a new development, a partial settlement somewhere else, a change in EU leadership, a competing complaint from a different sector, could dilute the momentum. Publishers who watched the AI Overviews rollout play out slowly in other markets over two years without triggering serious regulatory consequences clearly decided that speed, not patience, was the better strategy this time.

What this means if you're not a French publisher

If your brand doesn't touch French media law, none of this looks directly relevant, and that's exactly the mistake worth avoiding. Regulatory pressure on AI search platforms doesn't stay contained to the jurisdiction that files first. The EU Commission's December 2025 investigation and this month's French complaint sit inside the same broader push, and California, the UK, and several other EU member states are watching the same playbook for signs it works before deciding whether to run their own version.

That watching-and-waiting posture is exactly what happened previously with the neighboring-rights framework itself: one country established the legal mechanism, a handful of early adopters tested it against a slower-moving platform, and once it produced an enforceable result, the approach spread more broadly. There's no reason to expect AI content licensing to follow a meaningfully different pattern this time, and every reason to expect it to move faster, given how much more visible and how much more precisely measured the AI traffic-loss numbers already are compared to the earlier rounds of platform-versus-publisher disputes.

For B2B brands and publishers watching from outside the news industry, the practical takeaway isn't legal, it's strategic. The traffic decline AI Overviews cause isn't a temporary bug regulators will quietly fix on your behalf. It's the reason being the cited source inside the answer matters more than ranking below it, because a citation survives the traffic-loss trend that's driving this entire fight, and a ranking that no longer gets clicked doesn't. Whatever France's regulator ultimately decides, the underlying economics that produced this complaint in the first place, less traffic per search, more of the answer delivered inside the AI feature itself, aren't going away any time soon. Building for citation instead of just ranking is the strategy that holds up regardless of how any of these individual licensing fights resolve.

See where you are cited today

A free snapshot audit of your rankings and AI citations before we ever talk.