CIA Builds its Own AI Chatbot

The Central Intelligence Agency is on the road to release their own internal AI chatbot similar to OpenAI’s ChatGPT to help the agency’s data analysts find information based on their vast archive.

The U.S agency did not clarify on what type of model it will be using in its new tool. It also did not disclose how they plan to safeguard their information from leaks and prevent sensitive information seeping into the internet. It has most probably trained its bot with loads of data from its archive to make it understand what type of information analysts are dealing with.  

The U.S department of Defense aims to take advantage of the power of AI as it struggles to process vast amounts of data fast, and wants to use AI to speed up the process. 

“We’ve gone from newspapers and radio, to newspapers and television, to newspapers and cable television, to basic internet, to big data, and it just keeps going,” Randy Nixon, director of open-source division, CIA, told Bloombeg.

This tool allows agents to see the original source of information that they are viewing–to make sure that AI is not ‘hallucinating’ and the information is factually correct. 

The feature will be available to 18 U.S. intelligence agencies, which includes National Security Agency, FBI, CIA and branches run by the military. However, in the interest of national security, it won’t be available to the public or government officials. 

The U.S government has been exploring the potential uses of LLMs. They had set-up a task force earlier in May. “We need to find a  way to take advantage of these large models without violating privacy,” Gilbert Herrera, director of research at NSA, told Bloomberg.

This announcement comes at a time when China has publicly stated that it is harnessing the power of AI and plans to dominate space by 2030.