Hub, Spoke, and Code: Fighting Algorithmic Collusion in the AI Era

Written by Anjali Singh, the author is a law student currently pursing BBA.LLB from Symbiosis Law School, Pune

Introduction

With Artificial Intelligence taking the world by storm, it warrants all-comprehensive evaluation of legislations to ensure contemporary compliances. In the realm of competitive practices, AI poses new challenges in the form of AI based pricing algorithms and enforcement. The Competition Commission of India Chief at the 10th National Conference on Economics of Competition Law confirmed these concerns, highlighting that AI-based pricing might form “cartels without human communication” and “price coordination without explicit agreements”. Scholars have characterized this as ‘collusion by code,’ a problem that traditional competition law has difficulty addressing. 

Recent enforcement actions such as the U.S. Department of Justice’s antitrust lawsuit  against RealPage Inc. for engaging in anti-competitive rent pricing by landlords using its revenue management software and similar cases in the European Union (“EU”) and India pose regulatory concerns. The author seeks to shed light on algorithmic pricing collusion, the legal stance in India, the U.S. and EU and suggests potential steps that can be taken. 

What is Algorithmic Pricing Collusion? 

Algorithmic Pricing Collusion does not operate the way traditional cartels do. Traditional cartels require a “meeting of minds”, a handshake, a phone call, or at least a nod across a boardroom table. Modern “collusion by code” bypasses this entirely. Competitors feed profit-maximisation objectives into autonomous pricing agents, which then independently converge on similar prices through repeated market observation, no explicit instruction, no shared spreadsheet. Regulators are left chasing “intent” where none was ever recorded, because the intent was baked into the objective function at the point of design.

India’s Stance: No Meeting of Minds? No Cartel?

In 2018, CCI dismissed allegations of price fixing and hub and spoke arrangement against Ola and Uber. The Commission noted that algorithmic determination of fare setting did not add up to a cartel as the essentials like existence of a third-party platform and the intention to collude were absent. Further, there was no agreement between the drivers. Dynamic surge pricing is so far legal under India’s law, although standalone e-commerce regulations are contentious.

Further in Re: Alleged Cartelization in the Airlines Industry, the CCI examined supposed collusion by airlines on airfares using a shared pricing software. The CCI noted that the revenue teams input data into the software but final pricing decisions were made by officials. Thus, “the algorithm had merely facilitated price discovery and did not determine prices without personalised input,” therefore no collusion was established.

By and large, the Ola-Uber case was replicated in the airlines case: no algorithm-only cartel, as human managers still had control over pricing. The CCI closed investigation, suggesting that mere adoption of similar software, in the absence of proof of an understanding, was not enough for a Section 3 violation.

Antitrust Scrutiny in the United States and European Union

The U.S. Department of Justice (“U.S. DOJ”) and several State Attorney Generals filed a civil antitrust lawsuit against RealPage Inc. for reducing competition and propagating anti-competitive pricing via its revenue software which allowed competing landlords to exchange competitively sensitive information and coordinate rents. RealPage’s ‘recommendations’ allegedly encouraged landlords to ‘raise rents’ instead of compete. The Justice Department clarified that “utilizing software as the means of sharing does not immunize this scheme from Sherman Act liability”.

RealPage thereby shows a genuine “hub-and-spoke” model made possible by an algorithm, landlords (spokes), who supply data to RealPage (the hub) and use its price outputs. In early 2025, the U.S. DOJ secured a settlement with one of the defendants prohibiting unsupervised use of revenue management software, highlighting that under U.S. law, algorithmic price collusion is no different from old-fashioned cartels.

In a similar fashion, the EU also undertakes a heavy-handed approach towards algorithmic collusion. In June 2025, the German Federal Cartel Office (“FCO”) initiated proceedings against Amazon. The FCO expressed concern about “algorithmic mechanisms being used to influence third-party seller pricing.” In 2024, the European Commission had published initial opinions that Amazon might have infringed EU rules by promoting its own products on the platform, evidencing how platform algorithms can skew competition. The EU’s 2023 Horizontal Guidelines actively caution that jointly using pricing software or inputs “may effectively lead to parallel pricing”.

Analysis

Algorithmic collusion brings with it enforcement challenges, especially when coordination can take place without human interference. Scholars explain how “like-minded” algorithms will soon find it “easier to tacitly collude” upon learning market circumstances.

In India mere utilization of a shared algorithm or data does not attract liability under Section 3 of the Competition Act, 2002. Existence of an ‘explicit’ agreement or a coordinated scheme is essential. In the US, however, the RealPage lawsuit highlights one such avoidance, landlords specifically agreed to exchange data with RealPage, forming a clear agreement. But, if competitors merely license comparable software independently without explicit meetings, enforcement agencies will need to resort to more covert theories such as inference of common intent.

In hindsight, algorithmic collusion between competitors may be within the purview of Section 3(3) if it constitutes a horizontal agreement. The Competition (Amendment) Act 2023 in India even broadens the definition of “cartel” to purposely include hub‑and‑spoke structures, a recognition of digital-era collusion. Also, existing antitrust provisions are used by the EU and US agencies in algorithmic cases by treating the software as an intermediary, not a shield.

Yet, regulation of self-learning algorithms with no foundational agreement hangs in the air. The proposed U.S. legislation, Preventing Algorithmic Collusion Act introduced in 2024 intends to criminalise the use of algorithms to collude.  The Indian Digital Competition Bill, was weighing up ex-ante regulation of dominant online platforms although draft rules at present are more concerned with gatekeeper obligations than algorithms for prices and currently has been withdrawn adding to the uncertainty.

The Path Forward: Adapting Antitrust to the AI Age

While issues loom large, some steps can be taken to address the gap. Firstly, India might need to clarify whether coordination through algorithms is an “agreement” under Section 3. The Competition (Amendment) Act, 2023 is a step in the right direction, but stringent provisions banning “algorithmic pricing setting,” need incorporation as suggested in the U.S. legislation. Enforcement agencies must view algorithmic pricing software on the same thresholds as cartels. The RealPage lawsuit shows that sharing input data for generating recommended prices is per se unlawful.

India’s current stance is reactive instead of practicing proactive measures, waiting for harm to materialise before even beginning an inquiry. The CCI would do well to study France’s approach, which is focused at stress-testing pricing algorithms before they reach the market. Firms operating in India, by contrast, face no comparable pre-deployment scrutiny.

Conclusion

The CCI is powerless to combat twenty-first century AI cartels using twentieth-century laws that require demonstrating intent by a human agent. Ola/Uber and the Airlines inquiry both ended without result and not because there is no coordination at all, merely because the coordination is algorithmic and Indian law is searching in vain to find a handshake that will never be found.

Without the Competition Act amended to include the code as conduct, algorithmic hubs will still be operating unequivocally to orchestrate market prices. India has now a decision to make, work towards rebuttable presumption of coordination where competing firms implement functionally identical pricing system and observe correlated price changes, or wait until the damage is irrefutable and the market is already distorted. The debate has ceased to be on a yes or no basis whether algorithms conspire, it now extends to question, Whether Indian law can change at a pace sufficiently rapid to overtake them.