ANALYSIS: Antitrust Bills Aim at AI Pricing Collusion

Recent bills introduced in the Senate aim to augment the current antitrust laws by calling out algorithmic collusion—which can be greatly enhanced by AI.

It’s part of a broader attempt by antitrust enforcers and their allies in Congress to get ahead of the incoming market impacts of artificial intelligence before potential harms are baked in. And recent experience with algorithmic pricing suggests that some harms are pretty easy to predict.

Smoke-filled Digital Rooms

Gone are the old-school days of colluding with competitors in a dark room full of cigar smoke. As alleged in recent cases claiming that algorithmic collusion fixed prices in residential real estate, it’s possible with a pricing algorithm to share sensitive commercial information, agree on price levels, and police cheating on the cartel without ever meeting. The algorithm can synthesize the relevant data and “suggest” prices independently to competitors that coordinate across businesses seamlessly.

And that makes digital cartels challenging to prove under current antitrust law for several reasons. Collusion allegations require plausible proof of an actual agreement, and without meetings that’s more challenging for plaintiffs to show. If there is no plausible proof of agreement, claims can’t clear even the first hurdle to bringing a case.

Furthermore, because antitrust only punishes explicit agreements, not tacit agreements that might push prices higher just like an explicit one, the chances of a successful claim are longer when an algorithm does the dirty work behind the scenes. But because AI-powered algorithms are so adept at pulling together information, drawing conclusions, and making suggestions, lawmakers are bracing for more effective cartels facilitated in digital spaces with fewer overt signs of collusion (aside from stubbornly higher, uniform prices), which means they may be very harmful and very hard to detect and punish.

Closing Loopholes

At least two bills have been recently introduced to close these “loopholes” in antitrust law in anticipation of AI turbocharging the efficacy of pricing algos. Senator Amy Klobuchar (D-Minn.) penned the Preventing Algorithmic Collusion Act (S-3686), and Senator Ron Wyden (D-Ore.) introduced a broader bill (S-3692) aimed specifically at the residential real estate industry.

The text of Klobuchar’s bill isn’t posted yet, but it’s been read and introduced. As described, the bill would ban companies from using competitively sensitive information from their direct competitors to inform or train a pricing algorithm, and would create a presumption of illegal collusion if competitors violate that ban.

Wyden’s bill takes a slightly different tack. The bill would:

• Make it unlawful for rental property owners to contract for the services of a company that coordinates rental housing prices and supply information, and designate such arrangements a per se violation of the Sherman Act;
• Prohibit the practice of coordinating price, supply, and other rental housing information among two or more rental property owners;
• Make it unlawful for two or more coordinators to merge where a merger creates an appreciable risk of materially lessening competition; and
• Allow individual plaintiffs to invalidate any pre-dispute arbitration agreement or pre-dispute joint action waiver that would prevent them from bringing a suit under this act.

In other words, Wyden’s bill makes the contract with an aggregator in commercial real estate per se illegal, while Klobuchar’s would presume a violation if competitors provide sensitive information to an aggregator under that contract. Wyden’s bill also invalidates arbitration agreements that keep antitrust claims out of court or prohibit class actions, and deems evidence of tacit collusion to be “plausible” evidence of a conspiracy if it tends to show a coordinating agreement of the type forbidden in the bill.

Gathering Information

Klobuchar says that her bill also requires companies that use pricing algorithms to disclose that fact, and to permit antitrust agencies to audit the program. Combined with a mandate to the FTC to study algorithms in Klobuchar’s bill, the auditing feature has the potential to help antitrust regulators keep up with the development of AI pricing in real time.

For its part, the FTC isn’t waiting for a mandate to study the market: At a recent summit on tech and AI, the FTC announced an inquiry into partnerships and funding in AI. It issued orders to the companies involved in several prominent AI partnerships: Microsoft, Open AI, Amazon, Google, and Anthropic.

What’s Next?

Anything in the current legislative environment faces long odds of passage. However, both of these bills bear watching, as do others that target algorithms operating behind the scenes of platforms and markets.

Even if these bills never get a floor vote, they demonstrate an understanding of the legal challenges to bringing antitrust claims against AI-facilitated collusion, and a willingness to evolve antitrust statutes to meet new forms of collusion in the AI age.

Bloomberg Law subscribers can find related content on our In Focus: Artificial Intelligence (AI) page, and our Antitrust Practice Center.

If you’re reading this on the Bloomberg Terminal, please run BLAW OUT <GO> in order to access the hyperlinked content or click here to view the web version of this article.