AI must be a feedback loop

Silent security applications create siloes

While artificial intelligence largely reached the mainstream as a consumer-centric application, it’s becoming more entity-centric and bespoke to businesses – Darktrace's AI-based security being a perfect example.

Joshua Whyte of Darktrace, speaking at this week's Cybersecurity Festival in London, pointed out that existing security applications - be they threat intelligence, penetration testing, attack surface management (ASM) or something else - often fail to talk to each other, creating a siloed environment that does not scale well.

Whyte described a system Darktrace calls ‘the Cyber AI Loop', using outputs from each system to create inputs for the others - allowing ASM to talk to pen test and pen test to talk to threat intelligence. Such an approach has the added benefit of creating more data, which can help to improve AI systems.

Darktrace's vision is to present all of these systems, and cyber risks, in a single pane of glass.

"This will allow teams to focus time where it matters, and feeds back into our Detect and Respond features to implement countermeasures around vulnerable assets and pathways."