Google Search Chief Admits Preferential Treatment for Gemini AI in Antitrust Hearing Testimony

Google's Preferential AI Treatment Exposed: Search Chief's Admissions Bolster Antitrust Case

Damaging Testimony Reveals Gemini's Exclusive Access to Premium Search Features

In a pivotal moment for the tech industry's competitive landscape, Google's head of search Elizabeth "Liz" Reed delivered testimony on May 7 that legal experts are calling a significant setback for the search giant's defense against antitrust remedies. During cross-examination in a Washington courtroom, Reed made a series of admissions that confirmed what regulators have long suspected: Google continues to provide its proprietary AI service with exclusive advantages unavailable to competitors—potentially cementing the case for far-reaching structural remedies.

"The testimony effectively confirms what many in the industry have long suspected but couldn't prove," said a senior antitrust analyst who requested anonymity due to ongoing consulting relationships. "Google created a facade of openness with its Vertex AI program while maintaining a two-tier marketplace where its own products retain unique advantages."

The Gemini Advantage: "Additional Results" That Matter

The most consequential admission came when Reed acknowledged under oath that Gemini, Google's flagship AI service, receives "additional search results" including access to knowledge graphs and information boxes that competing AI systems cannot obtain—even when those competitors pay for Google's Vertex AI "traceability" service launched in 2024.

This disclosure cuts to the heart of the Department of Justice's argument about Google's continued anticompetitive practices following Judge Amit Mehta's landmark ruling last year that Google illegally monopolized the online search market.

"When you have the person who oversees Google's search business admitting under oath that they're still playing favorites with their own products, it's difficult to see how the company can continue arguing that current market conditions are fair or open," remarked a former regulator who now advises institutional investors on tech policy.

Reed's testimony revealed that while Google has opened basic search capabilities to some AI companies, these third parties receive fundamentally different—and less comprehensive—data compared to what Gemini obtains internally. Companies like Anthropic, Meta, and OpenAI must not only make do with this limited data access but must pay Google Cloud for the privilege.

The 2,000-Engineer Claim Unravels

Perhaps equally damaging was Reed's reluctant acknowledgment that Google's widely publicized claim about remedial compliance requiring the reassignment of 2,000 engineers—a figure the company has repeatedly cited in court filings and public statements—was merely a "rough estimate" developed in consultation with Google's legal team rather than through rigorous internal analysis.

"It was a number we discussed with our legal team," Reed conceded when pressed by government lawyers, adding that the figure had not been vetted through normal engineering channels.

This admission undermines Google's central argument that the DOJ's proposed remedies would be prohibitively disruptive to implement and could compromise user experience—a defense strategy that has helped technology companies fend off aggressive regulatory interventions in previous cases.

Privacy Concerns as Strategic Shield

Reed also characterized the DOJ's data-sharing proposal as "broad and invasive of privacy" that would give competitors a "treasure trove of data" potentially making them targets for hackers. However, when asked to quantify these risks or provide specific examples of how privacy safeguards might fail, Reed offered little substantive evidence.

"Google is deploying privacy concerns as a shield against competition," commented a data governance specialist familiar with the case. "The reality is that well-structured data sharing with appropriate auditing mechanisms can adequately protect user privacy while still enabling fair competition."

Reed further objected to another DOJ proposal that would grant websites greater authority to opt out of Google AI products, describing the level of detail required as "difficult to implement." This resistance suggests Google recognizes that greater website control could significantly impact its AI systems' training data advantage.

Market Reactions Signal Vulnerability

The financial markets have taken notice of these developments. Alphabet's stock declined approximately 8% following the combined impact of Reed's testimony and separate reports that Apple might switch Safari to a non-Google AI search provider—a dual threat to Google's core revenue streams.

"The testimony creates a perfect storm for Google investors," explained a portfolio manager at a major technology investment fund. "You now have both their primary distribution channel potentially at risk with Apple and their core competitive advantage in AI increasingly vulnerable to regulatory intervention. The market is repricing the company's growth prospects accordingly."

Reed's admissions substantially strengthen the DOJ's hand as Judge Mehta considers a range of potential remedies during this three-week hearing. Legal analysts now see several options gaining credibility:

Mandatory API parity requiring Google to provide identical Knowledge Graph and OneBox data to all qualified AI competitors
Royalty-free "grounding" licenses with usage tiers to prevent discrimination
Algorithmic firewalls prohibiting Search from favoring first-party properties
Structural separation potentially involving Chrome or Gemini divestiture if behavioral remedies are deemed insufficient

"Judge Mehta now has clear evidence that Google continues its monopolistic behavior even after being found liable," noted an attorney specializing in technology competition. "This shifts the calculus toward stronger structural remedies rather than mere behavioral constraints."

Historical Precedents Point to Strong Interventions

The current proceedings have significant parallels to earlier landmark antitrust cases. The Microsoft settlement of 2001 required interface disclosure and API parity—conceptually similar to equal-access search data requirements now being considered. More dramatically, the AT&T divestiture of 1982 demonstrated that structural splits become necessary when conduct rules prove insufficient.

"What we're seeing is reminiscent of AT&T, where the company fought against structural remedies by claiming they would harm consumers, only to see a post-breakup flourishing of the telecommunications market," observed a legal historian who studies antitrust enforcement. "Judge Mehta is likely considering whether Google's continued resistance justifies a similar approach."

Competitive Landscape Poised for Transformation

The implications extend well beyond Google. Meta, Anthropic, and OpenAI have already intensified lobbying efforts for fuller data access, sensing an opportunity to level the playing field against Google's AI dominance.

"This could trigger the most significant competitive realignment in search and AI since Google's original rise to dominance," predicted a venture capitalist who specializes in AI startups. "If the court mandates feature-level parity, we'll see a wave of new entrants challenging established players with innovative approaches previously blocked by data access limitations."

Smaller search startups, previously unable to compete against Google's data advantages, could find new opportunities in a post-remedy landscape—potentially triggering an innovation boom similar to what followed the AT&T breakup.

What's Next: Critical Timeline

The hearing continues through May 23, with additional Google witnesses and expert economists expected to testify. Post-trial briefs on remedy scope will follow in June, with Judge Mehta expected to issue a remedy order by August 2025. Appeals to the D.C. Circuit would likely extend into 2026.

For investors, four key takeaways emerge from Reed's testimony:

Reed's sworn statements fundamentally undermine Google's public claims that its AI marketplace is already "open and competitive"
Google's feasibility objections appear weaker after the exposure of unsubstantiated compliance cost estimates
Data-sharing mandates have moved from possible to probable outcomes, with significant implications for Gemini's competitive position
Structural breakup remains a viable remedy option, particularly if Google continues claiming that non-discrimination rules would "undermine user trust"

As the hearings progress, global regulators are watching closely, with European and Asian authorities likely to cite this testimony as evidence when crafting their own AI and search-gatekeeper regulations.

In just one afternoon of testimony, Liz Reed may have fundamentally altered the trajectory of both Google's business model and the broader competitive landscape for AI and search—providing market participants with clear signals that the era of Google's uncontested search advantages may be drawing to a close.