AI poses a dilemma for companies: Don't implement it; You may miss out on productivity and other potential benefits. But get it wrong; You can expose your business and customers to risks that cannot be underestimated. A new wave of “security” startups is coming in with the aim of making sure these threats, such as jailbreak and signal injection, cannot be ignored.
Like an Israeli startup Or and US-based competitors A hidden layer versus Protect the AI.British university spin-off Be careful One of these things. “AI is still software; So all the cyber risks you've heard about are AI-related,” says its CEO and CTO Professor Peter Garraghan (pictured right above). But “if you look at the transparent nature and intrinsically random behavior of neural networks and systems, it also justifies a new approach.
In Mindgard's case; The approach is Dynamic Application Security Testing for AI (DAST-AI), which targets vulnerabilities that can only be detected at runtime. It includes continuous and automatic red teams to map attacks based on Mindgard's threat library. For example, It can test the robustness of image classifiers. Opposite inputs.
Before and beyond that, Mindgard's technology comes from Garraghan's background as a professor and researcher focused on AI security. The field is changing at a rapid pace — ChatGPT didn't exist when I entered it. But he told TechCrunch that he feels NLP and image models face new threats.
From that time on, a voice that looked to the future emerged. It has become a reality. Within a rapidly growing sector, But LLMs are changing, as are the threats. Garraghan thinks his ongoing relationship with Lancaster University could help sustain the company: Mindgard will automatically own the IP for the work of 18 more postdoctoral researchers over the next few years. “There is no other company in the world that has such a contract.”
Despite its links to research, Mindgard is already a highly commercialized product. It is led by co-founder Steve Street as COO and CRO. (Early co-founder Neeraj Suri, who was involved on the research side, is no longer with the company.)
Enterprises are a natural customer for Mindgard, like traditional red teams and pen testers, but the company is also working with AI startups that need to demonstrate AI risk protection to their customers, Garraghan said.
Many of these potential customers are based in the United States, so the company added some American flavor to its packaging table. After raising 3 million pounds of seeds By 2023, Mindgard has now raised a new $8 million round led by Boston-based .406 Ventures, Atlantic Bridge; It is joined by WillowTree Investments and existing investors IQ Capital and Lakestar.
Boston-based Next DLP CMO Fergal Glynn, its recently appointed vice president of marketing, will fund “team building, product development, R&D and everything you'd expect from a startup” to expand into the US. However, the company plans to keep R&D and engineering in London.
With a head count of 15, Mindgard's team remains relatively small, with plans to reach 20 to 25 people by the end of next year. That's because AI security “hasn't reached its peak yet.” But when AI becomes ubiquitous and security threats follow, Mindgard will be ready. Garraghan said, “We built this company to do positive good for the world, and the positive good here is that people can use AI with confidence and safety.”