An Expectation Is Not an Agreement: Judge Mehta Tosses the Chegg and Penske Suits Over Google's AI Overviews

The first major test of whether antitrust law can hold a search engine accountable for AI-generated answers has ended in a dismissal. On Wednesday, October 1, 2026, US District Judge Amit Mehta of the District of Columbia threw out the consolidated lawsuits filed by education-technology company Chegg and Penske Media Corporation — the publisher of Rolling Stone, Variety, Billboard, and The Hollywood Reporter — which accused Google of unlawfully co-opting their content for AI Overviews and starving their websites of traffic in the process.

The ruling, reported first by Reuters and confirmed by The Verge, closes — for now — the most closely watched legal attempt to frame AI summarization as an antitrust problem rather than a copyright or legislative one. And its logic reaches well beyond these two plaintiffs.

What the case was about

The lawsuits, filed in early 2025, rested on a simple observation: when Google’s AI Overviews answer a query directly in the results page, many users never click through to the source websites. Chegg, whose entire business model depends on students visiting its homework-help pages, said the effect was existential — the company had already reported steep revenue declines and was exploring a sale when it sued. Penske alleged that traffic to its affiliate-linked stories collapsed, and that organic affiliate revenue fell by more than a third from its peak to the end of 2024.

Both plaintiffs advanced two core theories. First, a “bargain” theory: for decades, publishers let Google crawl their content for free in implicit exchange for search referrals. AI Overviews broke that bargain, they argued, by consuming the content without sending the visits. Second, a tying theory: publishers could not opt out of AI Overviews without risking their placement in ordinary search results, effectively forcing them to accept one product to keep access to another.

Mehta rejected both. “Plaintiffs have pleaded only that they have an ‘expectation’ that Google will send them search traffic if they make their content available for free,” he wrote. “But an expectation is not an agreement. It is simply how a general search engine works.”

The reasoning, in detail

On the bargain theory, the court found no plausible formal or implied contract. There was no communication establishing that the parties mutually agreed to exchange content for traffic, and no essential terms — no price, no quantity, no guaranteed volume of visits. Penske tried to infer a contract from the parties’ historical dealings, but Mehta noted that allowing content to be crawled at a price of zero does not constitute a negotiated bargain. He found it more plausible that publishers accepted free crawling because they independently wanted the traffic.

The judge also stress-tested the theory’s implications. If an implied contract existed between Google and every indexed publisher, Google would have agreements with billions of websites — a result the court was not prepared to accept. Even Google’s ranking incentives for higher-quality content, he reasoned, did not transform search visibility into a contractual commitment.

On the tying theory, the obstacle was product definition. Penske’s own account of user behavior — that a reader satisfied by an Overview has little reason to continue to another website — undercut the claim that traditional links and AI summaries were two distinct products serving distinct demands. Mehta viewed general search and AI Overviews as one integrated search experience, both answering the same underlying demand for information. Without two separate products, there can be no unlawful tie.

Harm acknowledged, remedy denied

Perhaps the most consequential passage is what Mehta did not do. He explicitly declined to dismiss the publishers’ predicament, writing that he was not “unsympathetic” to journalists, educators, and other creators whose work is repurposed without compensation. But antitrust law, he held, is not a substitute for legislation when innovation causes economic harm. The boundary he drew — between recognizing harm and finding the legal elements required to remedy it — effectively routes the publishers’ broader grievance to Congress.

The complaint’s damages figures remain allegations, not findings. But the underlying dynamics are visible across the industry: Chegg’s decline became a boardroom shorthand for AI-driven traffic loss, and publishers from news groups to independent sites have reported the same click-through erosion since AI Overviews rolled out broadly in 2024.

Why this ruling matters

Three takeaways stand out for anyone watching the intersection of AI and platform regulation.

First, this is the first appellate-adjacent US ruling on the theory that AI summaries can be an antitrust violation, and it went squarely for the platform. The same judge, it is worth noting, previously rejected similar claims by another publisher in March — a pattern that now looks like settled doctrine at the district level.

Second, the dismissal narrows the realistic path for publishers to copyright and licensing fights — the arena where OpenAI, Anthropic, and Google are already negotiating content deals and defending training-data lawsuits — and to legislative proposals for levies or mandatory compensation schemes.

Third, the timing matters. The ruling lands the same week Google shipped Gemini 4 Argon, a frontier model purpose-built to autonomously find, validate, and patch critical software vulnerabilities, and as AI Mode — a full conversational search surface that goes beyond Overviews — opens to all users. If AI Overviews were legally frictionless, the fully agent-driven search experiences now being built face even less exposure under US antitrust law.

Google denied wrongdoing throughout and argued it has no obligation to index publishers’ content on the publishers’ preferred terms — a reminder that the alternative to the current bargain is not necessarily a better one. Chegg and Penske did not immediately respond to requests for comment; an appeal remains possible, and the plaintiffs’ underlying claims could resurface in different form.

For now, though, the message from Judge Mehta’s courtroom is stark: the decades-old handshake between publishers and search engines was never a contract, and in the United States, at least, the bill for AI search will not be settled under the Sherman Act.