In 2016, while running my research lab, I began studying cybersecurity issues within AI and machine learning systems. I was struck by how challenging it is to imagine AI and ML becoming as commonplace as operating systems or virtual machines. They’re everywhere, right? So, I posited that if AI is going to be behind everything, making life better and everything, it is also going to attract problems from everywhere – be it countries, groups or individuals. The more AI is integrated, the more risks we will face. What about the cybersecurity tools we have? It is possible that they will have serious difficulties in this future.
I spent the next four years engaging in research with a group of brilliant scientists and engineers with whom I was fortunate to work. We had this big data center in my lab where we could investigate deep problems. It turns out we were well aware of the security challenges within AI. And then we realized we needed to create something, a product, to solve this problem. But to get there, we needed solid technology and all the research we were doing. Since then, it has really exploded, in a good way. And we remain true to one of our core principles that researching difficult problems is essential to remaining competitive.
How has the company evolved since its launch?
About two or three years ago, after we made some real progress on the research side, we were able to say, “Okay, we’re going to build a company.” Mental Guard was founded in May 2022. I laid all the groundwork to raise the initial capital, get the proposition up and running, move the technology through demos, etc. I basically worked two full-time jobs. This involved both operating a leading research laboratory in the UK and training the next generation of computer scientists, while carrying out much of the preparatory work for start-up over the next two years. I did this because I believe we really need to get our technology out into the world ahead of the growing threats to AI. I was due to come to London non-stop from Lancaster.
Tell me about the company: what it is, what its goals are, who you work with, how you reach customers, etc. ?
On one side, we have the decision-makers and on the other, the technical experts. Techies, including data scientists and cybersecurity professionals, often lack a deep understanding of AI security. Data scientists focus on creating and applying machine learning, without necessarily worrying about security, while cybersecurity specialists may feel overwhelmed by AI and not fully grasp its implications. We often find ourselves helping them understand the importance and specifics of AI security.
A year ago, our conversations might have seemed more “future-proof” to them, intriguing but not urgent. Six months later, there has been recognition of the need to prepare for AI security challenges. Today, our contribution is immediately requested as the projects are about to go live. We explain to technical specialists that despite the novelty of AI and LLMs, security management principles, such as threat detection, data protection and risk assessment, still apply. You just need to adapt them to the AI context.
Tell us about the work culture at Mindgard
I would say it’s really exciting. Everyone in the company is fully aware that we are working on unsolved problems with world-changing consequences. It motivates us in everything we do. When we create new technology, we help businesses understand the cyber risks of AI, while supporting and mentoring each other. Everyone is passionate about Mindgard’s greater purpose of securing AI on a global scale, and I am immensely proud of all my colleagues for what we have accomplished so far and will continue to do so in the future. future.
How are you financed?
As a spin-off from Lancaster University, we originally funded a lot of our initial research through grants etc. We continue to maintain a good presence of our activities in the academic field, due to the enormous amount of innovation.
On the business side, we raised capital in August 2023, from IQ Capital and Lakestar. Together we announced a £3m funding round in September.
The goal for the near future is to continue to rapidly grow our business, bring our products to market, and get this solution into the hands of as many people as possible to address real AI security challenges. This is essentially about expanding our focused team into a much larger entity. We are currently building a solid foundation; we simply need more resources to expand our reach by doubling down on product development and commercial deployment.
Many companies are on a similar journey, trying to come to terms with the deployment of AI, LLMs and GenAI. We anticipate that our fundraising efforts will align well with this change, putting us just ahead of the curve.
What has been your biggest challenge so far and how did you overcome it?
Many businesses are unaware of the cyber risks associated with AI. It’s extremely difficult for non-specialists to understand how AI actually works, let alone what the security implications are for their business. I spend a lot of time demystifying AI security into a clear proposition. At the end of the day, AI is still mostly software and data running on hardware.
You don’t need to throw away your existing cybersecurity processes, playbooks and tools, just update or re-arm them for AI/GenAI/LLM. This thinking has rapidly changed in recent months with the rise of GenerativeAI and LLMs, government attention, and working groups such as the OWASP AI Exchange that are truly helping to actualize and anchor cybersecurity risks within of AI.
How does Mindgard fill an unmet need?
To date, there is no major security platform for AI/GenAI/LLM. This presents a fundamental problem for companies looking to adopt AI (whether built in-house or through a third-party application), as they are forced to do everything security-related by hand. They likely realize very quickly that they don’t have the people or skills needed to keep pace with a rapidly changing field. Mindgard addresses this by providing a comprehensive AI security solution, including automated penetration testing, threat detection, response and data loss prevention (DLP), so businesses can use the AI in complete safety.
What does the future hold for us?
Two things: The first is that the field of AI will continue to innovate, allowing us to create things that seem, even to technically minded people, to be borderline science fiction. The second is that as AI adoption accelerates in the coming years, cyberattacks against AI will only increase in number and damage. This is exactly why Mindgard was created and is well-positioned to help people safely anticipate the tidal wave of AI.
What advice would you give to other founders or future founders?
It’s a long road, so I suggest you kindly get your motives in order. I always want to change the world. That’s why I started this company. The goal is to spread this technology to actually solve real-world AI and machine learning security problems.
In general, I left my job as a teacher, which I love, and which I still practice. It’s a very similar skill set and similar motivation in that there is a greater purpose in what I do – I’m here to solve problems and define the future, because I I can see in the next few years, a huge amount of AI security issues arise regularly. AI is evolving so quickly every week. Keeping pace and keeping up with that pace will be a huge undertaking. And if anyone is going to do it, it’s going to be us. We live and breathe this stuff.
And finally, a more personal question! What is your daily routine and rules you follow at the moment?
As a CEO and CTO, my routine is quite varied in terms of daily activities. I am also retaining my professorship at Lancaster to develop the next generation of AI security researchers in the UK. As you can imagine (like all startups), we are rather busy! A guiding principle I stumbled upon very early in my career was to “identify and focus on the things that matter most” – whether that be day-to-day operations or the strategic direction of the company .
Peter Garraghan is the CEO and co-founder of Mindgard.
