Knostic raises $16m to secure generative AI data

A Herndon cybersecurity startup creating fine-grained access controls to prevent data leaks secured approximately
$16 million this spring. Knostic, founded in 2023, provides need-to-know AI tools for companies using generative artificial intelligence large language models like Microsoft 365 Copilot.

“We have a lot of organizations rolling out AI systems to help knowledge workers be more productive,” says Knostic co-founder and Chief Technology Officer Sounil Yu. “But these AI systems can’t keep a secret.”

In March, less than a year after the start of a pre-seed round, Knostic, which has about 30 employees, announced it had raised $11 million in funding. Bright Pixel Capital, a Portugal-based venture capital firm that focuses on cybersecurity software, led the funding round. Silicon Valley CISO Investments and California’s DNX Ventures joined.

In April, the company won a $5 million simple agreement for future equity investment as a Top 10 finalist in the RSAC Innovation Sandbox, a competition that awards “cyber-security’s boldest new innovators.” The investment brings Knostic’s total funding now to $19 million.

In the next five years, Knostic plans to use the funding to continue developing access controls to get ahead of the explosive AI market.

“It’s been a wild ride because of the pace of both the adoption of the technologies that we’re riding on top of, as well as the introduction of new opportunities with the usage of AI,” Yu says.

Naren Ramakrishnan, engineering professor and director of Virginia Tech’s Sanghani Center for Artificial Intelligence and Data Analytics, says security controls are important with the rapid growth of generative AI.

“You can’t just walk into an office and take any file from any shelf that you want,” Ramakrishnan explains. “You need to have a key.” With AI systems, access controls provide that security.

Generative AI systems are eager to provide answers. Access controls, like Knostic’s, prevent the dissemination of confidential information like salaries and sales revenue, Ramakrishnan says.

Knostic’s software applies rules for what an employee should or should not see when they log on based on their role, Yu says.
“When businesses adopt these AI systems, they end up shooting themselves in the foot by exposing company secrets,” Yu says. “What we’re doing helps ensure that LLMs can exercise discretion.”

The rapid adoption of AI provides Knostic a space to persist through an ever-changing field, Yu says: “It’s pretty fundamental to help drive the next evolution of technologies.”