Comment from Fusion Collective

Fusion Collective respectfully submits this comment in response to the Commission's request for public input on the proposed policy statement concerning the suppression of accuracy in artificial intelligence systems, published July 1, 2026. This comment is limited to a single, discrete recommendation: the disclosure standard the proposed statement relies on to distinguish lawful conduct from a Section 5 violation is not administrable as drafted, and the Commission should adopt an objective, testable standard before finalizing the policy statement. The proposed statement holds that an AI company can avoid Section 5 liability by making "clear, conspicuous, and adequate disclosures" that its system prioritizes certain objectives over what a user requests or expects. The statement does not define that standard for a conversational AI output, and it does not adapt the term from any existing Commission framework built for this medium. This gap matters. The Commission has spent more than two decades refining an objective, multi-factor test for "clear and conspicuous" in digital advertising, first in the 2000 Dot Com Disclosures guidance and then in the 2013 .com Disclosures update. That test evaluates proximity and placement, prominence, distracting elements, repetition, duration, and plain language, against the net impression of the whole communication. Nothing comparable exists yet for a system that generates a novel, unscripted output in response to every prompt. Without an adapted standard, the meaning of "adequate" will be set case by case, after the fact, by whichever enforcement actions the Commission brings. That uncertainty falls hardest on smaller AI companies, since only the largest labs can absorb the cost of litigating an undefined standard. A vague disclosure exception does not reduce regulatory burden; it moves the burden from rulemaking to enforcement, where it is harder to anticipate and more expensive to contest. Fusion Collective recommends the Commission adopt, or issue companion guidance adapting, its existing clear-and-conspicuous factors for conversational AI outputs, addressing at minimum: •Proximity and placement: whether the disclosure is attached to the specific output it qualifies, not only in a term of service or one-time onboarding screen. •Prominence: whether the disclosure is set off from surrounding conversational text rather than folded into a tone-matched response. •Unavoidability: whether a user encounters the disclosure at the point where the objective actually shapes an output, not only once at account setup. •Repetition: whether a disclosure made once, months earlier, can reasonably inform a user's understanding of a specific answer today. •Understandable language: whether the disclosure states the actual tradeoff in plain language, rather than generic boilerplate about "balancing perspectives." Fusion Collective further recommends that the Commission not rely on self-attestation to determine whether a disclosure meets this standard. The consent orders the proposed statement itself cites, against Workado, DoNotPay, and Intellivision Technologies, each involved a company's own representations about its AI system's accuracy, representations the company believed were adequate and the Commission later found were not. The Commission should treat independent, third-party verification against a recognized AI governance standard, such as ISO/IEC 42001 or similar, as a mitigating factor, and where appropriate a rebuttable presumption of compliance, rather than accepting company self-certification at face value. Should the Commission decline to narrow or abandon this theory, Fusion Collective urges that it not proceed without first curing the arbitrariness the current draft invites. If the Commission proceeds regardless of the foregoing objections, it should do so only by adopting a defined standard for the disclosure safe harbor rather than leaving "adequate disclosure" to be settled case by case in enforcement. At minimum, the Commission should (1) adapt its existing clear-and-conspicuous factors; proximity, prominence, unavoidability, repetition, and plain language, to conversational AI outputs; (2) clarify that the disclosure obligation attaches where an undisclosed objective actually changes an output, not merely at the level of general marketing; (3) recognize independent, third-party verification against a recognized AI-governance standard as a mitigating factor or rebuttable presumption of compliance, rather than accepting company self-attestation; and (4) establish an advisory-opinion process so companies and their auditors can test a compliance posture before it is tested in enforcement. These measures do not resolve the foundational defects identified above, but they would at least ensure that a company acting in good faith can know in advance what the standard requires.