According to Bloomberg, Microsoft has launched a large language model (LLM) AI that operates completely separate from the internet, meeting the needs of analyzing classified information of US intelligence agencies such as the FBI and CIA.
This event marks an important step forward in the application of AI in the field of intelligence. The new AI model, based on GPT-4 and deployed in the cloud with an isolated system, allows US intelligence agencies to harness the power of AI technology to analyze classified information securely and effective. Previously, most AI models, including OpenAI's ChatGPT, relied on cloud services for learning and inference, posing a potential risk of data leakage.
Microsoft's development of an AI model that doesn't need the internet meets the urgent needs of the intelligence community amid the growing volume of classified information. Spy agencies around the world want to apply AI to analyze data, but are concerned about information security issues.
“This is the first time we have had an isolated version – meaning it is not connected to the Internet and is on a special network that only the US government can access,” William Chappell, director technology in charge of Microsoft's strategic missions and technologies, shares.
The GPT-4 model implemented is a static model, capable of reading files but not learning from them or from the internet. This ensures information security and prevents confidential data from entering the platform. It is expected that about 10,000 people will have access to this AI.
Microsoft's new AI service will undergo testing and validation by the intelligence community before officially going into operation. This event promises to open a new era for the intelligence field, with the ability to analyze information more quickly and effectively, contributing to ensuring national security.
However, the use of AI in intelligence also raises some ethical concerns. Some experts fear AI could be used to manipulate public opinion, spread false information or violate personal privacy. Therefore, the use of AI needs to be closely monitored and ensure compliance with strict ethical rules.
