AI Briefing: How Google and DOJ’s latest antitrust arguments address AI
Lawyers for the Department of Justice and Google filed lengthy written arguments Nov. 4 in the ad-tech antitrust trial ahead of closing arguments later this month. Totaling more than 1,000 pages, the revised findings of fact (FOF) threaded together court evidence and witness testimony along with each party’s argument for the case through a legal lens. One aspect is the role of AI in arguments from both parties, which has consistently evolved.
Recent innovations like large language models aren’t the focus of ongoing search and ad-tech trials between the DOJ and Google. However, experts have said court rulings in both cases could affect Google’s ability to amass data for generative AI like it’s done in the past. The latest filings also offer an updated look at both sides’ arguments and how they’ve evolved in the weeks since the trial began in September — with AI as just one narrow angle of viewing the lengthy legal text.
Even without mentioning newer areas of AI like LLMs or AI-generated content, both sides do talk about other major online platforms invested in AI like TikTok, Reddit and Meta. AI arguments in the revised versions also don’t drastically affect either’s arguments, said Sarah Oh Lam, senior fellow at Technology Policy Institute.
“The plaintiff has the burden to establish the relevant markets and evidence of harms caused by actions in those markets,” Oh Lam told Digiday via email. “… To the extent that building up a two-sided marketplace and improving it and growing it over time is a business and technological effort, the use of AI does not seem to be dispositive at least in either of the party’s arguments.”
Beyond the trial itself, others say competitors — including OpenAI, Perplexity and Amazon — could still dent Google’s dominance even if it takes a while. Adam Brotman, co-founder of the AI marketing tech startup Forum3, said generative AI platforms could help shift how consumers and advertisers use search. Brotman, who was previously chief digital officer at Starbucks, said it’ll be important to see how other giants like Apple roll out AI platforms like Apple Intelligence: “I would not bet against that being a big consumer behavior driver.”
“The entirety of this ad tech world is all about marketers trying to get people to buy their stuff,” Brotman said. “That’s the landscape that we’re going into, and I feel like we’re in the eye of the storm. It’s kind of quiet, and we’ll see what happens.”
How AI comes up in the DOJ’s case
In its pre-trial findings of fact, the DOJ argued that Google’s algorithms benefit from the scale of the company’s publisher and advertiser base, extensive transaction data, and search traffic. Those factors all provide an indirect network effect, which allows for better AI training and more cost efficiencies. DOJ lawyers also illustrated their points with courtroom testimony from trial witness like Index Exchange founder Andrew Casale and PubMatic co-founder Rajeev Goel.
“Seeing and winning more impressions enables ad tech products to utilize more data to improve the quality of their algorithms in real time, effectively improving the quality of their products,” DOJ lawyers wrote. “… For example, ad tech products can use data to create more accurate algorithms that help their publisher customers determine how to optimally set reserve prices for different demand sources.”
The DOJ also mentioned the role of the search case in which the court rejected Google’s argument that AI has already eroded its barriers to entry from rivals and startups. It also noted the court’s opinion that generative AI likely wouldn’t change the market dynamic for the “foreseeable future.” According to the DOJ, Google’s data advantages from the scale of its ad tech stack “will continue to benefit Google’s artificial intelligence going forward and serve as a barrier to entry for firms that lack such data.”
“The theoretical possibility that new forms of competition could diminish a defendant’s market position at some indeterminate future time cannot refute evidence of monopoly power today,” the DOJ wrote. “And here the facts show that Google’s monopoly power has been remarkably durable, and nothing in the foreseeable future threatens that durable market power.”
Google’s touts AI’s impact on innovation and competition
Google’s focus on AI has evolved from addressing its innovative role to talking about the tech as a disruptive force in the world of advertising. For example, Google’s revised FOF mentioned AI is reshaping the competitive landscape as ad tech rivals build their own offerings.
Algorithms play a pivotal role in AI tools that help improve advertiser results, according to Google’s filing, which also mentions them as part of its defense against government allegations for secret code-named projects like Poirot. Company lawyers also say the DOJ’s defined markets and category assertions don’t “capture the world of competitive pressures within ad tech” or innovations over time. It gave the example of how Performance Max helps advertisers automatically shift ad spend without needing to make decisions about formats or channels.
“By defining markets in tools that transact one particular form of advertising, Plaintiffs also ignore the impact of artificial intelligence,” Google lawyers wrote. “Plaintiffs have not only excluded much of the innovation and evolution in display advertising, but also selectively focused their case on only a set of products in that broader history.”
Google also used testimony from DOJ witnesses to illustrate its arguments for how AI has infused more competition in the sector while also addressing DOJ claims of Google’s dominance curbing competitors’ ability to compete without the same scaled data.
In its revised FOF, lawyers cited ad tech execs like The Trade Desk chief revenue officer Jed Dederick, who testified in court that TTD was building AI tools to automate the assessment of metadata when determining when advertisers should place ad bids. Another example Google gave was Criteo, which defense lawyers noted touts its ability to amass targeting data without operating digital platforms in a way that gives it advantages against major walled gardens.
“Both large and small ad tech providers are able to use transaction data to run experiments and improve their products,” Google’s lawyers wrote. “Even smaller ad tech providers can still achieve the necessary sample size to run experiments by operating with different percentages of a company’s daily data or over a longer period of time.”
Google also mentioned AI to illustrate the argument of it having a higher quality platform than competitors. For example, lawyers wrote about machine learning tools used to screen ”trillions of ads through Google’s systems” while general purpose AI tools help train new detection and threat prevention capabilities.
“Enforcing brand safety controls requires properly classifying publisher content and advertiser creatives,” Google wrote. “… When an ad transaction is facilitated by Google end-to-end, Google can pre-screen and classify both the ad and the content it is served.”
The examples also tie into Google’s reasons for not dealing with rival exchanges in some cases, which is party of its defense against what’s known as the “duty-to-deal” doctrine of antitrust law. However, as experts noted inside and outside the courtroom, Google’s efforts to stop brand safety with its ad tech tools hasn’t necessarily solved the problem or outperformed others.
Prompts and Products — AI news and announcements
- Amazon Prime Video launched a new feature called X-Ray Recaps, which helps people use generative AI to text recaps of shows.
- A federal judge dismissed an AI copyright lawsuit against OpenAI that had been filed by Raw Story and Alternet in February.
- The Washington Post launched a new generative AI search tool called “Ask The Post,” which uses LLMs to answer reader questions about various news topics.
- Perion debuted a new live CTV ad format called “Stay Live” that uses AI to integrate ads into pivotal moments of live sports.
- Perplexity AI developed a live election information hub to help users stay informed about results.
- OpenAI ditched the “GPT” in ChatGPT by switching to Chat.com for its new URL. The previous owner of the domain was HubSpot’s co-founder, who paid $15.5 million for it in 2023.
- U.S. intelligence officials said “Russian influence actors” were behind an AI video that spread false information about voter fraud in swing states.
- Speculation has begun over how a Trump presidency will affect U.S. efforts to create AI safeguards and other regulations around related areas like advertising, media and user privacy.
- Taco Bell parent Yum Brands touted AI efforts on its quarterly earnings call, which included using voice AI in drive-throughs to take more than 2 million orders. It’s also using AI in other areas like marketing, customer feedback and personalization.
Quotes from humans
“Today, despite an explosion in technology and worker productivity, many young people will have a worse standard of living than their parents. And many of them worry that Artificial Intelligence and robotics will make a bad situation even worse.” — U.S. Senator Bernie Sanders, in his statement about the presidential results.
Other stories from across Digiday
- How AI shaped the 2024 election: From ad strategy to voter sentiment analysis
- Presidential ads mark a divide as both candidates spend big in battleground states
- Inside Dow Jones’s AI governance strategy, with Ingrid Verschuren
- What a second Trump presidential term means for media and advertising
- Why the ad industry is redefining what it means to be a creator vs. influencer
More in Media Buying
Data licensing lawsuit adds a legal wrinkle to Omnicom’s planned acquisition of IPG
There’s been a lot of speculation about the value of Acxiom to Omnicom’s acquisition of IPG, but an ongoing court case over the data warehouse adds another layer.
Holding pattern: Omnicom, IPG and the deal that’s leaving marketers on edge
How Omnicom’s proposed acquisition of IPG keeps marketers guessing.
Here are the numbers to know in Omnicom’s potential purchase of IPG
The acquisition is expected to yield $750 million in annual cost synergies within two years.