FTC’s draft AI rules could force explanations for job & loan denials

What happened

On September 15th, the U.S. Federal Trade Commission (FTC) released a draft set of nationwide rules targeting automated decision systems. These systems encompass software that influences significant outcomes, ranging from screening resumes and scoring loan applications to flagging insurance risks, determining benefits eligibility, and monitoring employees. The FTC's proposal focuses not just on the AI model itself, but on the entire workflow, from data input to decision trigger.

The draft rules introduce several core ideas. These include explainability, requiring meaningful and understandable explanations for major automated decisions; a "human in the loop" mandate for high-stakes situations, allowing qualified human review and override; bias and impact audits to regularly test for unfair outcomes; and security by design for sensitive workflows. Additionally, the rules propose a right to contest decisions, ensuring clear paths to appeal or challenge outcomes made by these systems.

These rules are currently in draft form, with a public comment period open until October 26, 2026. Final rules are anticipated around mid-2027. The FTC's initiative aims to address the often-invisible influence of AI on modern life, where automated systems frequently make critical decisions without providing individuals with transparency or recourse.

Why it matters

The FTC's draft rules signal a significant shift in how automated decision systems are expected to operate, especially when they affect fundamental aspects of people's lives like employment, finances, and healthcare. By demanding explainability and human oversight, the FTC is pushing back against the "black box" nature of many current AI applications. This move aims to re-introduce accountability and transparency into systems that often operate without clear rationale or appeal mechanisms.

This regulatory direction impacts both the developers and users of AI. For companies deploying automated systems in areas like hiring, lending, or insurance, the rules would necessitate deeper scrutiny of their algorithms' internal workings and their external impacts. The requirement for bias and impact audits means a proactive approach to fairness, moving beyond just technical functionality to ethical and societal considerations. For individuals, this means a potential rebalancing of power, allowing them to understand and challenge decisions that can profoundly affect their livelihood and well-being.

The emphasis on human review and the right to contest decisions highlights a critical tension: the desire for efficiency and innovation versus the need for fairness and human dignity. These rules suggest that in high-stakes scenarios, algorithms should serve as advisors rather than final arbiters. This framework acknowledges that while AI can offer speed and scale, human judgment and the ability to correct errors remain essential when outcomes have life-changing consequences.

What to watch next

• Will the public comment period introduce significant changes to the draft rules, particularly regarding the scope of "high-stakes" decisions?
• How will companies that rely heavily on automated decision systems adapt their internal processes and technology to meet potential explainability and human-in-the-loop requirements?
• What will be the specific enforcement mechanisms and penalties outlined in the final rules, and how might they influence corporate compliance strategies?
• Will the final rules, expected mid-2027, create a precedent for similar regulations in other sectors or by other global regulatory bodies?

What this means for you

Business leaders and operators should view these draft rules as a clear signal of an evolving regulatory landscape for AI. It is prudent to begin auditing where automated decision systems are currently deployed within your organization, particularly in areas affecting hiring, employee monitoring, customer approvals, or any financial processes. Identify which systems make or significantly influence critical decisions about individuals.

For each identified system, conduct an internal review to determine if you can clearly explain its decisions to a human. If you cannot articulate "what factors mattered, and here's how" for a decision, your system may not meet future explainability standards. Proactively developing internal protocols for human review, appeal processes, and bias auditing now can position your organization to be compliant and responsible, regardless of the final rule wording. Treat explaining your AI as a normal business practice, not a future compliance burden.

Key takeaways

• The FTC released draft rules on September 15th for automated decision systems affecting jobs, loans, and other critical areas.
• The proposed rules require understandable explanations for AI-driven decisions, not just a vague "computer says no."
• Human review and the ability to override AI decisions would be mandatory for high-stakes situations.
• Organizations would need to conduct regular bias and impact audits for their automated systems.
• Individuals would gain the right to contest decisions and appeal outcomes made by AI.

What are the FTC's new draft rules about AI?

The U.S. Federal Trade Commission released a draft set of nationwide rules on September 15th concerning automated decision systems. These rules aim to regulate software that impacts significant outcomes like job applications, loan approvals, insurance risks, benefits eligibility, and workplace monitoring. The core ideas behind the draft include ensuring explainability of AI decisions, requiring human oversight for high-stakes situations, mandating bias and impact audits, and providing individuals with the right to contest decisions.

When will the FTC's AI rules be final?

The FTC's rules for automated decision systems are currently in a draft stage. There is an open public comment period that extends until October 26, 2026. Following this period, final rules are expected to be published around mid-2027. Until then, the proposed regulations remain subject to public input and potential revisions before they become official.

What is "explainability" in the context of these AI rules?

Explainability, according to the FTC's draft rules, means that if an automated system makes or drives a major decision about an individual, that person should receive a meaningful and understandable explanation. This goes beyond a simple "computer says no" to clarify "these factors mattered, and here's how." The goal is to provide transparency into the AI's decision-making process, allowing individuals to comprehend why a particular outcome occurred.

How do the FTC draft rules address human involvement in AI decisions?

For high-stakes situations, the FTC's draft rules propose a "human in the loop" requirement. This means that a qualified human must be able to review and override decisions made by an automated system. The principle is that while algorithms can advise, they should not be the final authority in critical matters. This ensures human oversight and intervention capabilities, particularly when decisions have significant consequences for individuals.

Can I appeal a decision made by an AI under these new rules?

Yes, the FTC's draft rules include a "right to contest decisions." This provision aims to establish clear ways for individuals to appeal or challenge outcomes determined by automated systems. If a system makes an error, the intent is that individuals would not be left without recourse but would have a defined process to seek review and potential correction of the decision.