Artificial intelligence: Collusion and competition

August 22, 2017

Figuratively speaking, “artificiality” has always been a feature of anti-competitive conduct since cartelists would meet to discuss confidential information with one another and then impact prices in the market, artificially affecting competition and ultimately harming consumers.

The Topkins  case[1], however, has raised regulator’s concerns regarding the use of artificial intelligence in a more literal sense. In 2015, Mr. Topkins pleaded guilty in a US court to involvement in collusion that manipulated the market and kept prices high for classic cinema posters sold on Amazon. While the subject-matter seems trivial, the method of price-fixing is inventive as it was conducted via a carefully crafted algorithm. The case was also investigated in the UK. 

The Topkins case and collusion using software tools presents competition authorities with another dimension to police which will inevitably present challenges for authorities.  The wide scope of on-line markets could also bring a risk of greater harm to markets than in the non-digital world.

While the longevity of cartels has always been subject to the human emotions of greed and fear of betrayal by a fellow colluder, the shift to “robo-sellers” (automated pricing robots) means that humans no longer need to be directly involved.  As a result, possible penalties, such as individual fines, sanctions, and liability, could no longer prove to be effective deterrents or incentives for whistleblowing programmes.

Not only could artificial intelligence increase the life-span of cartels, use of “robo-sellers” to collude in markets will require the authorities to find new ways to detect anti-competitive conduct. Efforts by competition authorities to detect and investigate collusion will require not only additional resources but will also require them to utilise similar artificial intelligence technology to “robo-sellers”. Ultimately and ironically, this might lead to an arms race between those using AI for collusive purposes and the authorities using AI to detect collusion and take action to protect competition.


[1] United States v. Topkins, U.S. District Court, Northern District of California, No. 15-cr-00201.