PII Comment from Dada, Muhammad Umair

FEDERAL AGENCY PUBLIC COMMENT Prepared by Muhammad Umair Dada Draft Comment for a Currently Open Federal Docket Subject: Policy Statement Concerning the Suppression of Accuracy in Artificial Intelligence Systems Issuing Agency: Federal Trade Commission (FTC) Federal Register Citation: 91 FR 41638, published July 7, 2026 (Document 2026-13628) Regulations.gov Docket ID: FTC-2026-0859 Matter Number: P264200 Dear Federal Trade Commission: I submit this comment on the FTC's Policy Statement Concerning the Suppression of Accuracy in Artificial Intelligence Systems (91 FR 41638, July 7, 2026, Docket FTC-2026-0859) as an individual software engineer whose engineering practice includes the design and integration of AI-augmented identity verification, facial recognition, and access-management platforms for U.S. digital services across healthcare, finance, transportation, education, retail, logistics, and the public sector. I hold a Master of Science in Computer Science from San Francisco Bay University and am the lead author of a peer-reviewed empirical study of U.S. software firms published in Remittances Review, Vol. 9, No. 4 (August 2024), pp. 4175 to 4189, ISSN 2059-6588. I offer three technical recommendations to strengthen the Policy Statement's engineering underpinnings. First, on the technical definition of accuracy suppression. The Policy Statement is correct that steering AI outputs toward undisclosed objectives, contrary to user expectations for accuracy, is deceptive within the meaning of Section 5 of the FTC Act. However, in engineering practice, accuracy is a measurable property that can be reported at multiple points in the AI pipeline: model precision and recall, calibration, false-positive and false-negative rates, and end-to-end system decision quality. I recommend that the Policy Statement clarify that a firm's disclosures under Section 5 should reference the specific accuracy metric being reported, and that firms should be required to disclose when the deployed system's operating point is not the model's Bayes-optimal or F1-maximizing point. This gives consumers a comparable technical basis for evaluating firm claims and reduces the ambiguity that today allows firms to advertise "best output possible" while operating at a suppressed accuracy point. Second, on the engineering audit trail required to detect accuracy suppression. Detection of accuracy suppression requires access to the deployed system's decision logs, threshold configuration, and post-deployment monitoring data. In my engineering work on identity-driven access management platforms (including a facial recognition system engineered to handle 100,000+ concurrent identities at 99.9% recognition accuracy), the audit trail includes decision records, confidence scores, threshold parameters, and post-decision reconciliation. I recommend that the Policy Statement encourage firms deploying AI systems in consumer-facing contexts to maintain audit trails that a subsequent Section 5 inquiry could examine to verify the reported accuracy metrics. NIST Special Publication 800-210 (General Access Control Guidance for Cloud Systems) and NIST Cybersecurity Framework 2.0 (February 2024) both provide baseline technical guidance on audit-trail integrity that this Policy Statement can reference. Third, on cross-cutting alignment with NIST AI RMF 1.0. The NIST AI Risk Management Framework 1.0 (January 2023) provides the federal government's principal technical guidance on AI system trustworthiness, including accuracy as a core trustworthy AI characteristic. I recommend that the Policy Statement expressly reference NIST AI RMF 1.0 and encourage firms to align their accuracy disclosures with the AI RMF's measurement and governance functions. This creates coherence across federal AI policy: firms that adopt the NIST AI RMF's measurement practices are simultaneously supported in meeting Section 5 accuracy-disclosure expectations, reducing the risk of accidental noncompliance and increasing the audit reliability of firm disclosures. These recommendations are drawn from my direct engineering practice on AI-augmented systems and from the peer-reviewed empirical evidence I have contributed to the U.S. software engineering literature. They are offered to strengthen the technical basis of the FTC's Policy Statement and to align it with the federal AI framework ecosystem developed by NIST under Executive Order 14179 and America's AI Action Plan. I appreciate the opportunity to submit this comment. I remain available to provide further technical input to the Commission on AI accuracy measurement, audit-trail engineering, and cross-framework alignment. Respectfully submitted, Muhammad Umair Dada | Eastvale, California |