FTC Proposes Mandatory Bias Disclosure Rules for AI Companies

The Federal Trade Commission has released a policy proposal that would compel artificial intelligence companies to disclose biases in their generative AI systems and large language models when those biases lead to inaccurate outputs.

The initiative, posted July 1, 2026, with a public comment period closing July 31, represents the agency's latest effort to apply consumer protection standards to AI technology. The FTC argues that consumers reasonably expect AI systems to provide accurate information, and any systematic deviation from truthfulness without clear disclosure constitutes a deceptive practice under the FTC Act.

What the proposal requires

The policy does not ban bias in AI systems outright. Instead, it mandates that AI companies inform users when their systems prioritize objectives that differ from what users would expect—particularly when those priorities compromise accuracy.

According to the proposal, if an AI maker allows or encourages its system to produce "less than truthful outputs," the company must disclose this tendency to users. The requirement applies specifically to generative AI and large language models, technologies that have faced persistent scrutiny over bias issues since OpenAI's ChatGPT launched in November 2022.

Enforcement context

This disclosure-focused initiative follows the FTC's ongoing campaign against false advertising by AI companies. The agency has previously targeted AI makers for misleading marketing claims and "over-the-top" advertising that misrepresents their systems' capabilities.

The new proposal shifts attention from how companies market AI to how the systems themselves operate and what users are told about their limitations. The FTC frames this as an extension of its consumer protection mandate rather than a departure into new regulatory territory.

Implementation challenges

The proposal leaves significant questions unresolved. Defining what constitutes "truthful" output in AI systems presents a complex challenge, as does determining effective disclosure methods that users will actually understand and act upon.

The initiative has already sparked debate about whether it represents necessary consumer protection or government overreach. The public comment period is expected to surface strong viewpoints on both sides, with some critics likely to characterize the effort as an "AI witch hunt."

Why it matters

This proposal could establish the first federal standard for transparency around AI system behavior, not just marketing claims. If implemented, it would require companies to acknowledge and communicate known limitations in their AI systems—a disclosure burden that could reshape how AI products are positioned and sold. The policy also signals that regulators view AI accuracy as a consumer protection issue, not merely a technical problem for companies to solve internally. How companies define and disclose "bias" under such rules could set precedents for AI governance globally.

The details were first reported by Lance Eliot writing for Forbes.