On Thursday, July 29, 2023, The United States Federal Trade Commission (“FTC”) issued a statement expressing their competition apprehensions over generative artificial intelligence (“AI”) technology. The FTC's statement, which we have summarized below, serves as a reminder that FTC is closely monitoring AI technology, and considering many related issues, including antitrust and privacy issues. Companies in the AI market should be mindful of how they design their ecosystems to help ensure such systems are not set up in a way that would run contrary to the FTC’s established rules.

Issues With Controlling More Than One Generative AI Building Block          

The FTC expressed concern that a scarcity of the three “key building blocks” that generative AI relies on – Data, Talent, and Computational Resources, may lead to a lack of competition in generative AI markets:

            (1) Data: Training a generative AI model from scratch to be at the level of a more well-known model like ChatGPT requires a large volume and quality of data. Even though it is not unlawful for a company to have a large amount of data, the FTC is concerned that the requirement of a large amount of data will keep “new players” from entering the market.

            (2) Talent: Developing and maintaining generative AI requires rare skillsets and deep understanding of machine learning, natural language processing, and computer vision. Due to the nature of the profession, it can be challenging to find, hire, and retain the talent required. Accordingly, the FTC suggests companies refrain from hindering their employees with non-competes because “it is critical that talented individuals with innovative ideas be permitted to move freely.”

            (3) Computational Resources: Not only does training an AI model from scratch require significant data and talent, it also requires substantial computational resources. This is because computational resources typically require specialized chips like graphical processing units (GPUs) that are expensive to operate and maintain. The FTC warns that the high cost of entry to creating a pre-trained base model may lead to a market where a small number of incumbents control the highest value models.

Open-Source Ecosystem May Not Really Be “Open”

While the open-source generative AI image ecosystem allows models to be accessed and used to develop new models by anyone, the FTC warns that these open-source AI models are vulnerable to misuse. Companies may use the “open first, closed later” tactic to undercut long-term competition by drawing in business while establishing a steady flow of data and then later close off their ecosystem to “lock in customers” and “lock out competition.” Once this occurs, companies will have a closed-off ecosystem and be able to control access to the generative AI blocks mentioned above while closing the gates to new players.

Possibility That Big Players May Use Unfair Methods of Competition 

The FTC warns that large scale companies already active in generative AI “may try to buy up critical applications and cut off rival access to core products,” effectively keeping the activity of generative AI in the hands of a few top players. Additionally, when referring to the AI market, the FTC stated that unfair competition may lead companies to participate in “discriminatory behavior.” Because certain companies could use their influence to gain an advantage in the marketplace, they could foreclose competition through various means such as bundling (offering multiple products together as a package), tying (conditioning the sale of one product on the purchase of another), or funneling users toward their own generative AI products rather than their competitors.

Network and Platform Effects Can Increase Harms

According to the FTC, companies engaged in generative AI could also “take advantage of network . . . and platform effects” to maintain a dominant position or concentrate market power. A large player could gain a competitive advantage as a result of its models interacting with a greater number of users over a longer period of time than its rivals’ models. A company could also leverage platform effects to their advantage, by exploiting the needs of customers and forcing them to be reliant on the company’s platform for their generative AI needs. The FTC references the cloud services space as an example, where cloud service providers could lock in customers by “charging exorbitant data egress fees.”