Everything seemed fine when one of Eran Barak’s customers, a financial services business, hired a remote worker to do their Salesforce admin. In the interview, the man seemed capable of doing what should’ve been routine work, and he passed all the background checks. Two weeks into the job, though, Barak says his AI software, dubbed MIND, spotted the employee sending highly sensitive company data up to his personal cloud account. The person coming into the office turned out not to be the same person who’d been interviewed, Barak says. “It was a completely different individual who had ‘bought’ the job from a professional fraudster.”

With so many new risks to business’ data, from accidentally employing a fraudster or losing data to North Korean spies with a similar modus operandi, or employees giving away company secrets to AI chatbots like ChatGPT, companies need new security tech to help them monitor and flag these risks.

Barak, who cofounded Seattle-based cyber startup MIND in 2023, said his company’s AI can prevent costly data leaks with an “autopilot” that autonomously identifies sensitive data and helps IT teams secure it, claiming it can reduce false alerts about data leaks “to nearly zero.”

“We want to push what we call a small language model.”

On Wednesday, MIND announced a $30 million Series A round led by Paladin Capital Group and Crosspoint Capital Partners at a $101 million valuation. That follows a $11 million round in September last year after it came out of stealth.

Barak says MIND’s AI is far better than a human at determining the risk of a given piece of data, whether that’s a batch of credit card numbers or meeting minutes, and then telling IT what layers of security are protecting them, if any. That means the AI won’t leave as many “blindspots” where sensitive data is left exposed, according to MIND’s pitch. To put these insights into action, an AI agent sits on each employee’s device and will detect and block data leakages either happening in the browser or via another app.

“We want to push what we call a small language model to the device itself, so we can really classify on the fly much more complex, sensitive data on the device,” Barak told Forbes.

The MIND cofounders have a history in successful AI startups. Barak cofounded Hexadite, another security automation tool that was sold to Microsoft for a reported $100 million in 2017. Hod Bin Noon, MIND’s VP of research and development, was a director at Dazz, which was acquired by cloud security giant Wiz for $450 million last year, with Wiz itself set to be acquired by Google for $32 billion. MIND CTO Itai Schwartz was a senior engineer at Torq, which makes AI agents focused on cybersecurity.

Barak “built one of the flagship security automation players,” said Paladin investor Gibb Witham, who led the round. He’d been on the hunt for an AI-native company to solve the problems modern AI was creating, Witham said, and the MIND team’s deep cybersecurity automation experience was a major draw.

Among MIND’s chief competitors is another startup, CyberHaven, which hit a $1 billion valuation earlier this year on a recent $100 million raise. Both are relying on AI to do much of the work of the IT security engineer, while looking for leaks across an organization’s networks.

Not that Barak thinks that AI is yet good enough to be relied on to automate all security tasks. “It won't completely replace us,” he adds. “But I can tell you that the depth and precision of the classification that we are doing thanks to the AI, it's unreal.”