The United States Congress has implemented strict restrictions on the use of generative artificial intelligence tools, such as ChatGPT, according to reports from Axios. The objective is to prevent the leakage of sensitive data and guarantee the responsible use of these technologies. The limitations affect congressional employees, who can only use the paid ChatGPT Plus service with more stringent privacy controls and exclusively for “research and evaluation.” In addition, access to non-public data is prohibited and manual activation of privacy features is required to prevent interactions from feeding the AI model. This policy seeks to prevent incidents such as the use of AI-generated ads for political purposes and the leakage of sensitive data. Below, we’ll explore the key details of this restriction and its implications.
Restrictions on the use of generative AI
Congressional Administrator Catherine Szpindor issued a memo detailing restrictions on the use of ChatGPT and other generative AI models in congressional offices. These restrictions seek to safeguard privacy and prevent possible abuse. Employees can only use the paid version, ChatGPT Plus, due to its higher privacy controls. However, its use is limited to “research and evaluation” activities, and cannot be part of the daily tasks of staff.
Even with the paid version, the use of ChatGPT is restricted to publicly accessible data. To prevent the collection of additional data, the privacy features must be activated manually. This prevents interactions with the chatbot from feeding the AI model. It should be noted that the free version of ChatGPT and other similar language models are not currently supported.
Concerns and justification
The implementation of these restrictions is not surprising, given that various institutions and companies have warned about the risks associated with the use of generative AI. In the past, Republicans have been criticized for using AI-generated ads, and leaks of sensitive data associated with Samsung’s use of ChatGPT have been reported. In academic settings, these systems have also been banned due to concerns of cheating and academic fraud. Therefore, the restrictions in Congress seek to avoid similar problems, such as the drafting of legislation and AI-generated speeches.
Regulating the use of AI in Congress
These restrictive congressional policies can receive broad support, as both houses of the legislature are working to regulate and oversee the use of AI. In the House of Representatives, Rep. Ritchie Torries has introduced a bill that would require warnings on the use of generative AI, while Rep. Yvette Clark seeks to establish similar requirements for political ads. Senators have also held hearings on AI and have proposed a bill to hold AI developers accountable for harmful content produced through their platforms.
The implementation of restrictions on the use of generative AI in the United States Congress is an important step to address the challenges and risks associated with these technologies. As AI continues to advance, it is essential to put in place proper regulations and safeguards to ensure its ethical and responsible use. While these restrictions may limit the capabilities of congressional employees, they also provide protection against potential data breaches and abuse. Privacy and information security must be priorities, especially in government settings.
