Harnessing AI for a safer digital world was a lead panel at day two of London Tech Week 2025.
Leaders in the AI space spoke about how businesses can prepare themselves against cyber threats and how AI can be utilised effectively in helping to do so; whilst being aware of its risks.
“The encouraging position of where we all sit today is what we do and how we build things really matters,” said David Stout, chief executive of WebAI. “We should not be building AI systems that take away from humanity, we should be building systems that’s accessible for everyone.”
“The most worrying things in cybersecurity are the things we haven’t thought of, because they are the things that will catch us by surprise,” added chief executive of the National Cyber Security Centre, Richard Horne.
The discussion examined the ‘attacker/defender’ approach to security, and how each of these sides work and operate in a cyber attack, made all the more relevant following a flurry of stories in recent weeks of major businesses in retail and beyond facing breaches.
“There is an asymmetry between the defender and the attacker,” said Darktrace’s Tim Bazalgette.
“The defender knows their own environment, they have an understanding that the attacker will never have. AI allows us to leverage and understand that environment better to use for defensive purposes.”

Feryal Clark, Parliamentary Under-Secretary of State at the Department for Science, Innovation and Technology, referred to how a vast majority of cyberattacks impact businesses with basic cyber vulnerabilities, despite AI making such attacks more “intense”.
“Far too many businesses do not prioritise cyber security […] that is the thing that keeps me up at night,” she added. Research has found that the majority of successful cyber attacks against businesses are done by exploiting human vulnerability rather than weaknesses in IT systems, making preparation all the harder.
Horne agreed that the foundation is key. “At the end of the day, the fancier we get with our technology, we are still building on the same technological foundations,” he added. “Unless these can be fixed, we are just leaving ourselves more exposed. The more we adopt technology, the more exposed we are to the impact of cyber-attacks and the more we therefore need to double down on our defences.”
Partnering with more bespoke AI models is key to businesses being cyber secure, according to Stout. “The worst thing that can happen is to adopt a tool that knows all the answers,” he explained. “You need to adopt a tool that solves a problem in the most esoteric way possible […] you need to find a model that is not going to convince your experts to change their minds.”
Read more: LTW 25: Peter Kyle addresses how avoiding risk can resist innovation
Register for Free
Bookmark your favorite posts, get daily updates, and enjoy an ad-reduced experience.
Already have an account? Log in