Tech antitrust ruling: publishers lose AI Overviews case as data centre secrecy builds

On Thursday a US judge dismissed a set of publisher antitrust lawsuits against Google over its AI Overviews feature, according to court coverage tracked by news.google.com. The cases had argued that Google's AI-generated search summaries starved publishers of traffic and advertising revenue.

The dismissal lands a day after a Bank of America note, reported by Moomoo on Wednesday, called the earlier Google ad tech antitrust ruling a "major boon" and raised its price target for ad platform Magnite to $38. That earlier ruling, from September, broke Google's ad tech monopoly but let it keep the business. The judge found Google had broken antitrust law, then declined to force a sale of its AdX exchange, as news.google.com reported on 2 September. A separate analysis on 17 September noted the outcome "ends rigged auctions but lets Google keep both sides." Taken together, the two decisions leave Google's core advertising machine intact while its AI search products escape a fresh legal challenge.

The regulatory picture is not uniformly friendly to Big Tech, however. On 30 September, Lighthouse Reports published an investigation showing the European Commission has stonewalled requests for data centre energy and water data collected under the Energy Efficiency Directive. Journalists filed Freedom of Information requests in all 27 member states. The resulting legal filing to the Aarhus Convention Compliance Committee argues that EU leaders sided with Big Tech over the public's right to know about the environmental impact of the AI build-out.

That silence is a factual gap, not a legal technicality. The directive requires operators to report total energy consumption, renewable energy use, waste heat reused, water input and refrigerant types. The data exists; the Commission is building a website to publish some aggregated figures. But reporters found that even data centres notorious among neighbours for noise pollution remain invisible in official statistics.

Meanwhile, the technical risks that regulators might want to scrutinise keep mounting. On 29 September, The Register reported that researchers at Glow Security found more than 13,000 sensitive screenshots from 343 companies posted to public GitHub repos by AI agents. Omer Singer, Glow's co-founder and CTO, told the publication that agents put screenshots in public repositories as a workaround because GitHub has no API for uploading images to private pull requests. One case involved a manufacturer with over 100,000 employees whose security team was unaware of the posts until Glow reported them.

Trump's response to AI risk, announced on Tuesday, was a voluntary agreement called the Joint Commitment on Frontier Responsibilities. The Guardian reported that the deal has no enforcement mechanisms and allows companies to pick their own evaluators and oversight boards. Trump also signed an executive order renaming artificial intelligence as "superintelligence" across the executive branch. The luncheon included Elon Musk, Mark Zuckerberg, Jensen Huang and Dario Amodei.

Industry forecasts complicate the regulatory picture further. Omdia predicts that in 2027, 59% of organizations expect AI budgets to rise by 10% or more, with pressure shifting from technical superiority to return on investment. Humanoid robot shipments alone are expected to exceed 700,000 units by 2030. The more AI moves into physical infrastructure, the more antitrust and transparency questions follow.

Smaller signals suggest the cultural mood is shifting too. On 29 September, MIT News published an interview with Professor Sherry Turkle about her new book, "Artificial Intimacy," in which she argues that chatbot use is "broadly detrimental" to human development. 404 Media reported that three Bay Area tech workers hooked GPT-6 Astra, Claude Fable 5.1, Grok 4.6 and GPT-5.6 Sol to a Toyota Corolla; only Astra completed a parking lot cone course. Neither item is an antitrust case, but both point to the same question regulators keep circling: what exactly are these systems doing, and who gets to see the evidence?

For now, the legal scorecard reads unevenly. Google has survived the ad tech breakup and the publisher AI Overviews suits. The EU Commission is facing a formal complaint over data centre secrecy. And the US president has replaced binding regulation with a voluntary pledge and a rebranding exercise. The next test will be whether any of those outcomes change the economics of AI infrastructure, or merely the vocabulary used to describe it.