Google Wins Dismissal of Publisher Antitrust Suits Over AI Overviews

The publishers had argued that AI Overviews, the AI-generated summaries that now sit above traditional search results, let Google keep traffic and ad revenue that would otherwise have flowed to news sites. The court rejected the claims. The ruling lands on the same day as a separate decision in a related matter, according to headlines circulating on Thursday 1 October.

It is the second antitrust defeat for Google's critics in under a month. In early September a judge found the company had broken antitrust law in its ad tech business, but declined to order the sale of its AdX exchange, letting Google keep the business intact. Bank of America analysts called that ruling a "major boon" and raised their target price on ad platform Magnite to $38, according to Moomoo. The ad tech case was never a clean sweep for either side. A widely shared analysis of the unsealed ruling noted that it ended rigged auctions but allowed Google to keep both sides of the ad market, meaning the structural problem at the heart of the complaint survived the verdict. That framing, published in September, is now the template for reading the AI Overviews dismissal: the court is willing to name the conduct, less willing to break up the company. Google's critics have had better luck elsewhere. The same week as the ad tech ruling, a surgical robot company told the Ninth Circuit that the decision "expands" antitrust law, according to Law360. The detail matters because it suggests courts are still feeling their way through how old competition rules apply to software and AI products, rather than settling the question.

None of this is happening in a vacuum. The same week the publisher suits were dismissed, Donald Trump announced what he called a "morally binding" agreement among tech CEOs to self-police artificial intelligence, and signed an executive order renaming AI as "superintelligence" in official documents, The Guardian reported on 30 September. Trump described the agreement as "almost like a constitution in a way" and said it would allow companies to perform "tremendous self-policing".

The document sets out four "layers of controls and audits", including internal safety monitoring during training, an internal team to check that monitoring, an external auditor with access to safety controls, and an independent board to review reports. None of the controls involve government regulators, and companies can pick their own evaluators, appoint their own boards and decide whether to publish results, according to The Guardian.

"It's almost like a constitution in a way, and the biggest people in the world signed that, and I signed it as president, and it really is a form of protection," Trump said, according to The Guardian.

That self-regulatory posture is the backdrop for the antitrust fights. If Washington will not write AI rules, courts become the main venue where the tech industry's competitive behaviour gets tested. The publisher suits were one such test, and they failed.

Across the Atlantic, the fight is about disclosure rather than break-up. Lighthouse Reports published a joint investigation on 30 September describing how EU leaders sided with Big Tech over the public's right to know the environmental impact of the AI build-out, forcing a legal challenge under the Aarhus Convention. Reporters filed Freedom of Information requests in all 27 member states for data centre energy and water metrics collected under the Energy Efficiency Directive, and say they were stonewalled. The dispute matters to the antitrust story because the same hyperscale data centres are the physical infrastructure behind products like AI Overviews. If the public cannot see how much power and water those facilities consume, the argument that AI search is cheap to run becomes harder to audit.

The technology itself keeps moving. The Register reported on 29 September that researchers at startup Glow Security found more than 13,000 sensitive screenshots of corporate software projects, from 343 companies, posted to public GitHub repositories by AI agents. Glow calls the finding PixelLeak. Omer Singer, the company's co-founder and CTO, told The Register that agents posted screenshots publicly as a workaround because GitHub has no API for uploading images to pull requests in private repositories.

Separately, a team of Bay Area engineers calling themselves DrivingBench hooked GPT-6 Astra, Claude Fable 5.1, Grok 4.6 and GPT-5.6 Sol up to a rented Toyota Corolla and ran a cone course in a parking lot, 404 Media reported on 29 September. Grok, Sol and Fable only drove a few metres; GPT-6 Astra completed the course after troubleshooting. Co-founder Aditya Ramabadran told 404 Media the point was not practicality but to create a benchmark for LLMs acting in the real world.

That is the tension the courts are being asked to referee: products that behave in ways their makers did not fully specify, built by companies that have lobbied against strict oversight and now operate under a voluntary White House framework with no enforcement mechanism.

Omdia published its 2027 outlook on 30 September, naming AI monetisation, supply chain disruption, digital sovereignty and physical AI as the four forces shaping the next phase of competition. The firm said 59% of organisations expect AI budgets to rise by 10% or more in 2027, but that enterprises will judge spending by return on investment rather than technical superiority. It also noted hardware delays are already affecting 60% of PC channel partners.

If that forecast holds, the antitrust questions get sharper, not softer. A company that already controls search distribution and the ad exchange beneath it, and is now folding AI answers into the same results page, faces a different kind of scrutiny once customers start asking what the AI spend actually returned.

Not every input in this debate is a court filing. MIT News published an interview on 29 September with professor Sherry Turkle around her new book, "Artificial Intimacy", in which she argues chatbots offer "pretend empathy" and that heavy use is broadly detrimental to human development. Her claim sits awkwardly beside the industry's push to put conversational AI in front of more users, including through search.

Meanwhile the infrastructure race continues. Sustainability Magazine reported on 30 September that Oregon startup Panthalassa, which raised $140m in a round led by Peter Thiel and reached a valuation near $1bn, is building floating wave-powered data centre nodes 85 metres long, cooled by seawater and connected via SpaceX's Starlink. The IEA projects the sector's energy consumption will rise 30% annually through to 2030, when AI is expected to account for 3% of global energy use.

MIT Technology Review said on 29 September that it will publish its 2026 list of Climate Tech Companies to Watch on 6 October, noting that Big Tech is "backpedaling on its climate ambitions, as companies race to build massive AI data centers". The UN has said the planet will tip past 1.5C of warming "likely within the next few years".

For Google, Thursday's dismissal is a reprieve, not a resolution. The publisher plaintiffs can appeal, and the ad tech remedies remain under review. The pattern so far is consistent: courts are willing to find that Google broke the law, and unwilling to dismantle the business that flows from it.