Congress Reportedly Restricts Staff Use of AI Models Such as ChatGPT

Congress has implemented strict limitations on the use of ChatGPT and similar generative AI tools. According to a memo from House of Representatives administrative chief Catherine Szpindor, congressional staff are permitted to use the paid ChatGPT Plus service only under specific conditions. This service is chosen for its enhanced privacy controls and can only be utilized for "research and evaluation" purposes—not for everyday tasks.

Furthermore, House offices are restricted to using the chatbot solely with publicly available data while ensuring that privacy features are manually activated to prevent information leaks into the AI model. The free version of ChatGPT and other large language models are prohibited.

This cautious approach aligns with warnings from various institutions about the risks associated with generative AI, including the potential for accidents and misuse. Notably, the use of AI-generated attack ads has drawn criticism, and incidents such as Samsung employees accidentally leaking sensitive data through ChatGPT highlight these concerns. Schools have also imposed bans on AI tools due to cheating apprehensions.

The House restrictions aim to mitigate risks like AI-generated legislation and speeches. Interestingly, this policy may receive broad support, as both parties in Congress are exploring ways to regulate AI technology. Representative Ritchie Torres has introduced a bill that calls for disclaimers regarding the use of generative AI, while Representative Yvette Clarke seeks similar transparency for political advertisements. Additionally, senators are holding hearings and proposing legislation to hold AI developers accountable for harmful outputs generated through their platforms.