AI Briefing: How the AI race could pay a role in Google’s antitrust trial

Some see Google’s efforts to talk about AI as a means of blurring the lines between market definitions and creating confusion between the open-web display advertising market outlined by the DOJ. However, the definition of these markets is a key factor in the case being made by the DOJ and the states.

Despite new AI startups that have brought recent competition to the market, some AI observers argue that nothing has diluted the giant’s dominance — especially with rising players OpenAI and Perplexity still reliant on search technology and search ads from Google and Microsoft. As of May 2024, Perplexity was still just 3% the size of ChatGPT and 0.4% of Google, according to a recent analyst report published by Datos, a Semrush company. The report also noted only 16% of search users used chatbots while 99% still relied on traditional search.

Thomas Höppner, a competition lawyer and partner at Hausfeld, Google’s claims of AI being a disruptive force are instead a “red herring.” In a recently published article published by Competition Policy International (CPI), Höppner wrote about how Google’s dominance could allow it to evade transparency and accountability for generative AI search platforms like AI Overviews, which risks letting the highest paying advertisers receive preferential treatment in answers.

“Generative AI does not challenge Google’s entrenched dominance across the entire ad-tech stack,” Höppner told Digiday. “Where AI is used to improve programmatic advertising, Google is in the lead position to deploy it. In fact, being one of the — if not the — leading providers of AI services across the board, if anything the rise of AI has further strengthened Google’s position in the ad-tech stack.”

If Google chooses to offer more AI-driven adtech tools for free, it could entice advertisers to rely less on agencies — especially small and medium sized businesses — noted Charles Benaiah, a former ad-tech exec who was previously a partner at a VC firm that invested in digital startups. Benaiah pointed as an example to Google’s recent acqui-hire of the founders of Character AI, a conversational AI startup. That could help power conversational AI ads while also limiting a major player in the market: “If this were any other deal, it would have been discussed for days or weeks,” said Benaiah. “To Google this is small. It was barely mentioned.”

“Google didn’t ‘buy’ Character. It paid Character upwards of $3 billion to ‘license’ a tech that no one else was using and hired all their key people,” Benaiah said. “Just like Microsoft ‘acquired’ Inflection AI in March for a $650 million license and hired their people and had Inflection’s CEO join Microsoft as evp of Microsoft AI.” 

Google’s AI-powered ad tools are also part of the picture. In its proposed pre-trial finding of fact filed last month, Google touted its Performance Max platform to illustrate how its ad tech has evolved and to detail its innovation efforts. Describing it as an “immensely popular” way to use AI for buying different types of ads, Google mentioned its ability to reach people across more than 3 million websites and apps.

Google also noted competitors with AI tools. One example is Criteo, which Google noted “has exposure to daily data for more than 750 million active consumers and billions of opportunities to connect to consumers each day.” Another example mentioned by Google was Microsoft’s AI tool that also is called Performance Max. Google’s deposition of Simon Whitcombe, vp of Meta’s global business group, also includes the ways Meta uses AI for advertising.

Advertising experts expressed their concern about a potential lack of transparency with AI-powered platforms like (Google’s) Performance Max. They also worry that Google’s ability to amass data will help the company to improve and scale its AI tools for advertisers and users faster than smaller rivals. 

The DOJ’s original complaint includes more than a dozen mentions of Google’s algorithm, which the government said allowed Google to have “self-preferential algorithms and tech restrictions.” Another example mentioned was Google using its scale to have a flywheel network effect, which gives Google an advantage when combined with more recent algorithmic bidding programs. The DOJ also mentioned Google tweaking its “Project Bernanke” algorithm as part of a plan to lock in publishers that consider using rival platforms’ innovative bidding tech. 

Both the search and ad-tech antitrust cases are interconnected in another way, noted Amplitude Field CTO Ted Sfikas. He also warned of what could happen if Google is allowed to have AI-related revenue sharing agreements (RSAs) similar to what it made with Apple, AT&T and Verizon. With AI as the next frontier, Sfikas worries how Google might use RSAs to stamp out competing AI search platforms. “They need to become the best AI model in the world, or they’re going to lose their shirt and they’ll slowly be picked apart.”

Along with his concerns about transparency, Sfikas mentioned Google’s ability to control the entire data process — from collection, storage, and analysis to monetization — creating a cycle of revenue through products like Cloud, Analytics, Ads, Looker and Search. 

“This is an analogy of Exxon owning the ability to mine oil, owning the gas station that sells you to oil, and then buying the car dealerships that drive the cars that use the oil,” Sfikas said. “That’s the perfect analogy. We don’t let Exxon do that. Why should we let Google run the currency of the internet, which is data.”

Despite addressing AI’s potential benefits, Google’s filings also try to portray it as a white hat actor by acknowledging the risks of AI. For example, the company said it’s “aware that a rising threat from harmful ads comes from deep fakes generated by artificial intelligence.”

“The number of ads that Google has blocked has increased considerably from 2020 to 2022 because automation has permitted bad actors to connect to users at scale and grow their abilities to reach users and make money with their fraudulent tools,” read Google’s filing. “Google understands that number will only keep rising with the advent of AI and machine learning.”

Prompts and Products — AI news news and announcements

• Qatar Airways debuted a new “AI Adventure” campaign that lets potential travelers place themselves inside a destination story.
• Alibaba is working on a new “AI sourcing agent” that will include features for conversational search, intelligent comparisons and proactive recommendations.
• Yoast released a new AI-powered SEO optimizer for users, which comes after other recent debuts for AI search tools from companies like Semrush and Hubspot.
• Multiple companies offered new AI tools for brands safety and copyright. Seekr released a new self-service AI enterprise platform, while DoubleVerify announced a new AI detection tool for political content and deepfakes. Meanwhile, YouTube is working on new tools to protect creators and artists from having their image and voice copied by AI.
• Anthropic launched a new set of enterprise capabilities for its Claude family of AI models.
• Google released its Gemini Live voice AI feature for Android users. It also announced plans to roll its Video Action Campaign tool into the AI-powered Demand Gen platform.
• Meta joined the steering committee of C2PA, the trade organization focused on crafting new standards for AI-generated content.
• Bundeskartellamt, the German competition authority, is investigating how Amazon’s algorithmic enforcement of price parity impacts third-party retailers’ ability to set prices on its marketplace. As part of the investigation, the authority has launched an online survey of 2,000 retailers to gauge the impact of Amazon’s price reviews.

AI-related reports from Digiday and beyond

• Social media platforms fuel AI ad rush with new creative tools (Digiday)
• While 39% of marketers surveyed by Gartner using generative AI for social content, 62% think proper labels will have a positive impact, according to a new report from Gartner. (Digiday)
• Why AI’s impact on elections might be overblown (MIT Tech Review)
• Several people on the board of NaNoWriMo stepped down after the organization seemed to greenlight the use of generative AI for the annual writing event. (NYT)
• OpenAI co-founder Illya Sutskever raised $1 billion in funding for a new AI safety startup. (Reuters)
• Canva plans to increase the price of its AI tools next year, by more than 300% in some cases. (The Verge)

1s and 0s — Research and reports

• The Center for News, Technology and Innovation published a new report about perceptions of AI use in news and journalism. It also published a second report about what defines news, journalism and journalists. 
• The Dataset Providers Alliance, a data licensing trade organization, released a new paper outlining its stance on how creators and companies should approach ethical AI data licensing. (Read more about it in this Wired story.)
• NewsGuard uncovered a massive network that uses AI and fake accounts to spread misinformation.