
The Office of Legislative Counsel (OLC) of the U.S. House of Representatives has seen an increase in draft bills prepared using AI. This was reported by Politico, citing eight current and former employees of the department.
According to them, the problem arises during the review stage. AI-generated documents often contain incorrect references to existing laws, inaccurate definitions, and formulations that do not always align with the author’s original intent.
Lawyers sometimes spend more time correcting such material than drafting it from scratch. Any inaccuracies can delay the adoption of initiatives or lead to legal disputes.
Congresswoman Norma Torres warned that over-reliance on generative AI increases the risk of errors and formulations that do not reflect the legislator’s intentions.
“This responsibility cannot be shifted to a chatbot,” she stated.
Official data does not clarify how AI has impacted the workload of Congress lawyers. In April 2025, OLC head Warren Burke attributed the increase in requests partly to overall legislative activity and a large number of amendments.
In 2024, experts from POPVOX noted that large language models have facilitated the creation of pseudo-bills, increasing the flow of documents for legal review.
Congress Continues AI Integration
Despite issues with AI drafts, the House of Representatives continues to incorporate the technology into its workflows. As stated by Aubrey Wilson, Director of Government Innovation at the POPVOX Foundation, work with Microsoft Copilot in the lower house continued in 2026, and staff training on the tool began in February.
She emphasized that artificial intelligence should be implemented as an auxiliary tool to enhance human labor, not as a replacement.
The OLC is also exploring the technology. In April of last year, Burke reported that the service, along with other legislative bodies, is considering potential applications of neural networks, as well as the associated benefits and risks.
Earlier, in August, journalists from the Guardian found traces of AI hallucinations in a report for the Australian government. The document contained references to non-existent scientific publications.
