Autonomous systems accelerate the discovery of bugs and vulnerabilities, leaving organisations with far less time to identify and remediate their weaknesses
Goal-directed deception by AI was largely theoretical. The AI models that recently hacked IT systems show that deception can be a route to finishing a hard task
Success in AI adoption will belong to organisations that recognise that top-down strategy will only create value when it’s matched by bottom-up innovation
AI tools are shaping how we access information. They often feel like reliable experts, capable of answering almost any question in seconds. But appearances can be deceptive
We’re asking autonomous systems to make complex operational decisions, without having a complete picture of what’s happening at the edge of the network
AI is a powerful force for processing data at speed, but it often lacks the contextual reasoning required for safety assurance or complex risk management