In more detail

The Report, which was published in late January, summarizes submissions the Bureau received in response to the public consultation on its Discussion Paper: Artificial intelligence and competition ("Discussion Paper") that ran from March to July 2024. The Bureau issued the Discussion Paper to deepen its understanding of competition in AI markets and how AI may affect competition, particularly in the context of the Bureau's enforcement areas, including mergers and monopolistic practices, cartels, and deceptive marketing practices. For more information about the Discussion Paper, see our previous alert "Competition Bureau Seeks Comments on the Effects of AI on Competition".

In total, the Bureau received 28 submissions in response to the Discussion Paper from a diverse set of stakeholders, ranging from global technology companies to industry associations to class action law firms to consumer advocacy groups. These submissions are available here.

Key Takeaways

The Bureau summarized four broad takeaways from the written submissions: (1) AI is evolving rapidly, with large amounts of research and product implementation occurring daily; (2) investment is key for AI technology to grow, but there is a balancing act required between incumbent firms and start-ups; (3) AI may lead to anti-competitive conduct that may require new or revised enforcement approaches; and (4) the Bureau is viewed as an active collaborator, with future consultation and collaboration ideas offered by stakeholders. Highlights from the Report include:

• The complexity of the AI market distinguishes it from other digital markets. While the AI market shares similarities with other digital markets, it is generally more complex due to higher marginal costs, the prevalence of more partnerships in the AI industry and easy applicability of AI to a wide range of sectors. AI markets also tend to involve a broad range of stakeholders and significant investments including by AI developers, end users and infrastructure providers. Key competition issues identified in the submissions included dependence on resources (e.g., compute power, data centers and specialized chips), control over data creating potential barriers to entry, new competitive dynamics (e.g., algorithmic pricing) and barriers to entry for new players. The Report does not define "AI markets" as a relevant product market for competition law purposes.
• New entrants may face barriers to entry as a small number of companies hold significant shares of certain AI input markets. Larger firms predominate in the AI landscape and often control data, compute power and human expertise due to the high costs associated with compute resources and access to extensive proprietary data. However, the Report notes that some submissions stated that smaller firms are not necessarily at a disadvantage. This was attributed, in part, to the growing availability of compute inputs via cloud and open-source computing, and the promotion of partnerships between larger and smaller firms.
• Vertical integration as a potential barrier to entry. The Report highlighted that multiple stakeholders took the view that vertical integration and partnership of larger companies operating across various levels of the AI technology stack may create barriers of entry. They noted that exclusive agreements between cloud infrastructure providers and major developers of large language models might lock out smaller firms from accessing key AI inputs and entering or expanding. These barriers could be lowered by safeguarding firms' access to data inputs, developing public policy that promotes investments in compute infrastructure, and promoting fair partnerships among firms of all sizes. However, certain stakeholder submissions countered that these barriers are, in fact, not as high as alleged. For example, these stakeholders note that many of the AI 'unicorns' did not rely on data inputs or proprietary data from large firms, and that market entry is becoming easier with the increasing availability of open-source AI technology and public cloud infrastructure.
• "AI democratization" is increasing market access for smaller enterprises. "AI democratization" is the process of making AI more accessible to smaller firms which can then use the technology to compete more effectively in a number of industries, diversifying markets and creating positive market effects (e.g., more consumer choice, additional price options and higher quality goods). By reducing traditional constraints such as financial resources, intellectual property and human resources, AI democratization may allow smaller enterprises to compete against larger incumbent firms. 
• Double-edged sword of vertical integration, partnerships and investments. Certain stakeholders highlighted the benefits of vertical mergers, partnerships and investments, which, they noted, can reduce costs and enhance efficiency. However, the Bureau also received submissions arguing that large-scale acquisitions may eliminate rivals, threaten access to key AI inputs, and enable surviving firms to coordinate through data aggregation, raising the concern that these practices could lead to monopolistic control in the AI sector, reducing competition and innovation. 
• Algorithmic pricing may create competition concerns. Certain submissions recognized the advantages of algorithmic pricing. For instance, a submission from the American Bar Association stated that algorithmic pricing might improve competition by allowing companies "to maintain a consistent pricing approach aligned with their goals while rapidly adjusting to market shifts". Other submissions, however, highlighted concerns (which have also been raised by US and UK competition authorities) contending that algorithmic pricing may facilitate tacit collusion in which systems autonomously align on prices even without explicit human instruction, communication, or agreement. Hub-and-spoke conspiracies which might be facilitated through AI-based algorithms in which a single algorithm serves as the central hub, with various competitors (the spokes) relying on it to determine their pricing, were a common concern in submissions. 
• Existing law may not address AI-driven collusion. The Report indicates that several submissions expressed concern that Canada's current competition laws are not "fully equipped to handle the complexities of AI-driven collusion". The Bureau acknowledged in the Report calls for transparency rules to prevent extensions of market power, but stated its preference that these be developed strategically so as to mitigate the lessening of competitive benefits that overly-inclusive transparency rules may create.
• AI and deceptive marketing. Over a third of the submissions to the Bureau expressed concern regarding AI's potential ability to be used to further deceptive marketing practices. Similar to the concerns expressed in the Bureau's annual Fraud Prevention Month campaign in March 2024, the Report indicates that submissions reiterated generative AI's growing and novel abilities to defraud customers (e.g., by generating fake online reviews, endorsements, impersonations, tailored phishing campaigns, deepfakes, etc.). Several submissions advocated for new AI-specific deceptive marketing enforcement measures, and endorsed "labelling" as a way to promote transparency and informed decision-making associated with AI in marketing materials. 
• Legislation should be technology-neutral. Submissions advocated that technology-neutral legislation based on universal principles such as transparency, consent, and non-discrimination was a better fit for an evolving and dynamic AI market, as compared to technology-specific legislation, which could "hinder innovation, impose difficulties for growth, and create barriers to entry". 
• The Bureau should embrace its new market study powers. Several submissions urged the Bureau to take advantage of its market study powers (which were expanded through recent amendments to the Competition Act) to better understand the AI technology stack and market conditions. Market studies are in-depth examinations of a market or industry by the Bureau to identify factors affecting competition and propose solutions. As part of these powers, the Bureau may seek an order from the Federal Court of Canada to require third parties to produce information and data relevant to the subject of a market study.
• Cooperation is crucial for effective regulation and enforcement in AI-driven markets. The Bureau heard that it should cooperate with its counterparts in the UK, France, the EU, and the U.S., including aligning policies, and sharing best practices and insights to ensure effective enforcement in AI markets. Various submissions also advocated for a clear definition of AI, but recognized that the Bureau would likely need to adapt the definition from draft federal AI legislation – Artificial Intelligence and Data Act – Canada ("AIDA") – that was pending adoption during the consultation period. However, AIDA "died" when Canada's Parliament was suspended in January 2025, and depending on the outcome of a federal election may not be re-introduced in the same form (or at all), leaving the question of how to define AI for purposes of competition law unresolved.

The Bureau has gained important insights from the consultation and will continue to deepen its understanding of AI markets and the impact of AI on competition in 2025, including testing its understanding in the course of investigations, merger review and other activities. This remains an area to watch closely as the Bureau and competition authorities around the world continue to assess the intersection of AI and competition.