AI-generated 'slop' introduces legal 'bugs' into US House drafting
Staffers and legal experts warn that AI-produced errors in bill text could create systemic vulnerabilities in the federal legal code.
Artificial intelligence is increasingly used to draft and review legislation in the US House of Representatives, but the resulting output is raising alarms over accuracy. Legal experts and congressional staffers warn that the proliferation of AI-generated content—often termed 'slop'—is introducing significant errors into the foundation of federal law.
According to reports from Politico, some congressional staffers and advocacy groups have integrated AI tools into the drafting process to increase efficiency. However, this has led to the production of low-quality text containing critical mistakes, including the incorrect citation of existing statutes. These errors often fail to accurately reflect the intended policy goals of lawmakers, requiring intensive human intervention to correct.
The risk of legal 'bugs'
The integration of Large Language Models (LLMs) into government processes was intended to streamline the comparison of bills and the creation of initial drafts. Yet, the inherent tendency of these models to hallucinate or misinterpret the rigid, complex structures of legal language has created a new vulnerability in the legislative pipeline.
Ari Hershowitz, a lawyer and technologist, has characterized the trend as a dangerous gamble with the precision of the law. Hershowitz warned that allowing AI-generated errors to slip into official legislation is a guaranteed way of introducing "bugs" into the US legal system, mirroring the way a small coding error can crash a complex software program.
Why precision matters
Legislative language serves as the definitive blueprint for federal law. Unlike a standard business document, a single misplaced word or an incorrect statutory reference in a bill can create massive legal loopholes, unintended consequences, or direct contradictions within the federal code. When AI produces 'slop' that is not rigorously vetted, the risk is not merely a typo, but a fundamental compromise of the reliability of the legal system.
If these errors go unnoticed during the fast-paced legislative process, they could lead to years of litigation to resolve ambiguities created by an algorithm rather than a deliberate policy decision.
The path forward
As the House continues to navigate the adoption of AI, the focus is shifting toward the necessity of human-in-the-loop verification. While the tools offer speed, current evidence suggests they cannot yet handle the nuance required for statutory drafting without significant risk. Observers are now watching whether Congress will implement formal guidelines or oversight mechanisms to ensure that AI-generated drafts are subjected to mandatory, expert legal review before they reach the floor for a vote.