Mandates an annual AI review of the Code of Federal Regulations to identify and eliminate outdated or redundant regulations, streamlining federal rules and processes.
Jon Husted
Senator
OH
The Leveraging Artificial Intelligence to Streamline the Code of Federal Regulations Act of 2025 mandates an annual review of federal regulations using AI to identify and eliminate outdated or redundant rules. This review is overseen by the Office of Management and Budget, utilizing AI systems that meet specific standards for accuracy and security. Agencies are required to act on the AI's findings by rescinding or updating regulations, with all determinations published for transparency. This act also amends existing law to expedite the removal or modification of these identified regulations.
Bill Orders AI to Annually Scan Federal Rulebook, Flagging Outdated Regs for Quick Removal
This bill proposes using artificial intelligence to give the massive Code of Federal Regulations an annual spring cleaning. Starting within 90 days of enactment, an AI system, meeting standards set by the National Institute of Standards and Technology (NIST), would scan federal rules each year, flagging regulations it deems 'outdated' or 'redundant' (Section 3). Think rules that duplicate others or haven't kept pace with technology. The goal? To trim the regulatory fat and potentially make government rules easier to navigate.
AI Takes a Red Pen to the Rulebook
Imagine an AI reading through all the federal rules – that's the core idea here. Under Section 3, the Office of Management and Budget would deploy an AI system, supposedly vetted for accuracy, transparency, and fairness according to NIST guidelines, to perform this yearly scan. When the AI flags a rule as potentially 'redundant' (basically, a duplicate where removal causes no loss) or 'outdated' (superseded by new laws or tech, like rules referencing floppy disks), it gets kicked over to the agency responsible for that rule. The bill requires annual reviews of the AI process itself to ensure it's working as intended.
Agency Call: Keep, Update, or Trash?
Here’s where the human element comes back in, but on a tight leash. Agencies get just 30 days to decide if they agree with the AI's assessment that a regulation is outdated or redundant (Section 3). Critically, the agency's determination is final. If they agree a rule is redundant, it gets scrapped within another 30 days. If it's deemed outdated, they have 30 days to either update it or remove it entirely. To facilitate this, Section 4 amends existing administrative law (Title 5, Section 553 of the U.S. Code) to allow this expedited process for AI-identified regulations, bypassing some standard notice-and-comment procedures.
Cutting Red Tape or Cutting Corners?
So, what does this mean in the real world? On paper, it could streamline things. Businesses might face fewer overlapping regulations, and rules could potentially become clearer if genuinely obsolete requirements are updated or removed. The government hopes for increased efficiency. However, there are significant questions. How reliable is the AI? While Section 3 mandates adherence to NIST standards, defining and ensuring 'accuracy' and 'fairness' when an AI interprets complex legal text is a major challenge. Could an AI mistakenly flag a crucial environmental or consumer protection rule as 'outdated' simply because the technology mentioned is old, even if the underlying principle remains vital? Furthermore, with agencies having the final, binding say within a short 30-day window (Section 3), is that sufficient time for a thorough review? Or could this expedited process (Section 4) inadvertently become a fast track for removing regulations deemed inconvenient, potentially reducing oversight or protections without robust public discussion?