In a submission to Ottawa’s consultation on AI transparency, the Competition Bureau reiterated that there is "no AI exception" to existing competition laws. While new transparency measures are being explored, businesses must ensure their current AI marketing claims, algorithmic pricing models and vendor disclosures comply with the Competition Act today, including the strict substantiation requirements for deceptive marketing. This guidance is essential for corporate counsel, marketing teams and executives responsible for procuring AI tools or promoting AI-enabled technologies.
The Competition Bureau (Bureau) has made a submission to Ottawa’s consultation on AI transparency. It recommends three principles: transparency should support well-functioning, competitive markets, obligations should be proportionate and preserve competition, and regulation should focus on harms rather than technologies. For businesses, the submission is a reminder that existing competition law obligations apply to AI-related claims and conduct, even as new transparency measures are considered.
The consultation, led by Innovation, Science and Economic Development Canada (ISED), ran from July 23 to Sept. 23, 2026. Its discussion paper, Enhancing trust in artificial intelligence through increased transparency, asks whether new transparency measures are needed in five areas: AI-generated content, AI interaction, information about AI systems, AI incidents and AI agents. The paper seeks input on possible measures rather than proposing a settled regulatory framework.
The Bureau’s submission, Advancing artificial intelligence transparency, is dated Sept. 23, 2026. Read alongside the Bureau’s broader statements on AI, it offers insight into its approach to transparency and competition. The submission does not create new legal obligations. Businesses should nevertheless assess how the Competition Act applies to their AI-related marketing, disclosures and use of AI systems.
Practical takeaways for businesses
The Bureau’s submission to Ottawa is policy advice, not a new AI enforcement regime. Its practical significance lies in its emphasis on existing law and on transparency measures that address harms without undermining competition.
Businesses developing, procuring or deploying AI should consider four priorities:
- substantiate public claims before making them;
- assess how AI systems affect pricing, marketing and customer interactions;
- retain supporting records and secure appropriate access to vendor information; and
- review whether transparency initiatives could disclose competitively sensitive information or facilitate coordination.
Technology-neutral rules remain the starting point
The Bureau’s submission favours a technology-neutral approach that focuses on the conduct and harms at issue, rather than the use of AI itself. In the Bureau’s words, there is “no ‘AI exception’ to the law.”
That position is consistent with the Bureau’s contribution to the Canadian Digital Regulators Forum’s joint May 2026 article. It described the Competition Act as a federal law of general application that applies across technologies and sectors, and stated that firms remain accountable for their conduct regardless of whether AI is involved.
Businesses should not treat the absence of an AI-specific Competition Act regime as a compliance gap. Existing rules on deceptive marketing, competitor collaboration, abuse of dominance and other conduct continue to apply. Whether a particular use of AI contravenes the Act depends on the conduct and the requirements of the applicable provision.
Transparency can help competition, but it has limits
The Bureau supports transparency where it helps consumers make informed choices and promotes accountability. Its May 2026 article points to information about when and how AI is being used, and about the types and sources of data used to train AI models, as examples of information that can support informed decision-making.
The Bureau’s submission to ISED is more nuanced than a general call for disclosure. Transparency obligations should be proportionate, and their design should account for costs and competitive effects. The Bureau warns that disclosure can expose confidential business information or “facilitate coordination among competitors.” It also cautions that prescriptive, costly or technically complex requirements may disproportionately burden smaller firms and new entrants while favouring incumbents with greater compliance resources.
That qualification matters, because a requirement to tell a customer that they are interacting with AI raises different competition concerns from a requirement to disclose detailed information about a pricing algorithm, training data or system logic. The relevant questions include what information is needed, who needs it, and whether public disclosure is necessary to achieve the objective. Businesses should also consider whether voluntary transparency initiatives involve the exchange of competitively sensitive information, including information about pricing tools or system logic.
Businesses should assess both the usefulness of a disclosure and its potential competitive effects.
Existing deceptive marketing rules already reach AI claims
Although the submission does not expressly discuss section 74.01 of the Competition Act, its observations about misleading AI outputs are relevant to businesses’ obligations under the existing deceptive marketing provisions.
AI raises deceptive marketing risks in a number of related contexts, including how AI products and services are marketed, and how AI is used to create or deliver representations to consumers. The submission notes that AI systems “can also be used to manipulate, mislead, or distort decision-making.” Existing deceptive marketing rules therefore apply both to claims about AI products and to representations made through AI-enabled tools, including chatbots, virtual assistants, recommendation systems, and personalized marketing.
Under the Competition Act, representations made to the public must not be false or misleading in a material respect, assessed by both their literal meaning and the general impression they convey. Public claims about a product’s performance, efficacy or length of life must also be based on adequate and proper testing conducted before the claim is made. The person making the claim bears the burden of proving that testing is adequate and proper.
Businesses should review AI-related claims against these requirements. Does a description such as “AI-powered,” “intelligent” or “autonomous” convey more than the system actually does? Is an accuracy or reliability claim supported by testing that reflects the product as deployed? Does a model or system card used in marketing create a broader impression than the underlying evidence supports?
Also, where a chatbot or other AI interface communicates with customers, businesses should assess whether its responses accurately describe products, prices, terms, availability and policies, and whether qualifications or limitations are presented clearly enough to avoid a misleading general impression. A business should not assume that a representation falls outside the deceptive marketing provisions merely because it was generated dynamically by an AI system, rather than drafted or reviewed by a person.
AI may also shape price representations. Algorithmic pricing tools can determine or adjust the prices, discounts or offers presented to consumers, including by using information about an individual consumer or segment - a practice sometimes described as surveillance pricing. These practices are not necessarily deceptive, but representations made in connection with them remain subject to the ordinary requirements applicable to price claims.
Businesses should assess the general impression created by personalized prices, reference prices, discounts and time-limited offers; whether mandatory fees are disclosed upfront; and whether consumers are misled about the basis, availability or relative advantage of an offer. Particular care may be required where a consumer is led to believe that a price is generally available, objectively determined, or uniquely discounted, when it is instead personalized or generated using behavioural or other consumer data.
The same issue arises with AI-generated content. The Bureau has identified fake reviews, endorsements and impersonations as examples of traditional deceptive marketing practices that AI can amplify. Its submission treats labelling as one tool, not a complete answer: labels alone “may not be sufficient,” and labelling measures should be “carefully designed, empirically tested, and potentially combined with other tools.” Whether disclosure is required will depend on the context, including whether the absence of disclosure creates a materially false or misleading general impression.
Labelling content as AI-generated does not, by itself, resolve whether a promotional representation is materially false or misleading. The label does not cure an inaccurate claim, an unsupported performance representation, a misleading omission, or a deceptive general impression. Businesses must still assess the substance, context and presentation of the representation, including the prominence and clarity of any qualifying disclosure, and retain support for claims that require substantiation.
Accountability across the AI value chain
The Bureau has also been direct about responsibility. In May, it stated that companies cannot rely on automation or technical complexity to shield themselves from liability, and that developers and deployers are accountable for outputs they control in the value chain.
AI systems often involve several parties: a model developer, a platform provider, an integrator and the business that deploys the system to customers. The Bureau’s statement emphasizes accountability for outputs within a firm’s control; it does not establish a standalone legal test for liability. Responsibility in a particular case will depend on the facts and the applicable Competition Act provision.
For businesses that buy rather than build AI, vendor arrangements can help manage these risks. A business may need information from its vendor to understand a system’s output, substantiate a public claim, or respond to a regulator. Businesses should consider contractual rights to documentation, testing information, logs, incident notification and technical support, with the scope calibrated to the system and the risks involved.
These rights warrant particular attention where AI is used in pricing, advertising, rankings, recommendations or direct customer interactions.
Substantiation and records will matter
Businesses should retain the evidence supporting their AI-related claims and understand what additional information they can obtain from vendors.
Relevant, non-privileged records may be compelled through court orders under section 11 of the Competition Act, where the statutory requirements are met. Businesses should identify who holds relevant records, and assess whether vendor arrangements provide access to the information needed to substantiate their claims and respond to an inquiry.
