A group of former British spy chiefs, including Sir Jeremy Fleming, former head of the Government Communications Headquarters (GCHQ), has launched a company aimed at protecting corporate secrets from being inadvertently exposed to artificial intelligence. The startup, AI Score, is designed to help businesses monitor and control the use of AI tools in the workplace, addressing concerns that chatbots such as ChatGPT could ingest confidential company data and potentially repeat it outside the organisation.
AI Score provides software that can be installed on corporate networks to track employee interactions with AI tools, scanning inputs for sensitive information and identifying which platforms are being used. The company is already working with a large law firm, a consumer business, and a fintech company. Its launch has been backed by investment from Gallos, a firm run by former intelligence personnel and advised by Sir Jeremy. Gallos was established in 2021 by former high-ranking intelligence official Josh Burch and armed forces veteran Dean Jones.
The company will be run by Alex Harland, a founding team member of GCHQ’s National Cyber Security Centre. Sir Jeremy Fleming and Josh Burch will sit on AI Score’s advisory board, alongside other former intelligence leaders, including Gaven Smith, ex-GCHQ technology chief, and Tom Hurd, a former homeland security advisor. The startup has already raised over $1 million in funding from angel investors and Gallos.
Josh Burch highlighted the urgency of addressing AI-related risks in the corporate sector, noting that “large parts of the corporate world haven’t yet absorbed the risks they’re running.” Gallos research suggests that 80 percent of corporate security leaders feel unable to monitor how AI is being used across their organisations.
AI Score’s technology is designed to flag potential breaches by monitoring employee inputs to chatbots, identifying trigger words and phrases that could indicate sensitive information is being shared. The issue has already drawn warnings from companies including Samsung and Amazon, which caution staff against entering confidential data into AI systems. The UK Cabinet Office similarly advises civil servants never to input classified or sensitive information into AI tools.
This is the second startup incubated by Gallos, which emphasises its deep connections within UK and US security networks. The firm claims its network allows access to “unique deal flow and highly expert human capital” and previously helped establish StirlingX, a drone technology company. With AI adoption accelerating rapidly across the corporate world, AI Score aims to provide organisations with the tools and expertise needed to safeguard sensitive data in the age of generative artificial intelligence.
