The United States House of Representatives has established new restrictions that forbid the use of large language models in its offices, with the sole exception of OpenAI’s ChatGPT Plus service. This is an important step towards protecting data privacy, the house said.
The top administrative officer of the US House, Catherine Szpindor, relayed this information in a memo that highlighted recently implemented limits and cited security concerns as the primary motivator.
House offices that use the ChatGPT artificial intelligence chatbot are only permitted to use the ChatGPT Plus version of the programme, according to Szpindor. In order to secure House data, the product’s Plus version includes crucial privacy protections, she said.
No additional ChatGPT or other large language models, AI software are currently approved for use in the House, according to the memo. The memo explains restrictions on how House members may use the programme.
The software can only be used for research and evaluation, thereby preventing its integration into routine workflow. The message further forbids the provision of any private data through prompts and demands that ChatGPT Plus be utilised with all privacy options enabled.
Users of ChatGPT and ChatGPT Plus have been able to delete their conversation history and accounts since April, thanks to OpenAI, but the removed information remains on the ChatGPT servers for an additional 30 days.