When AI Meets Cybersecurity: Challenges and Strategies for Professionals

The speaker, an adviser who works with vulnerability management teams, opens by explaining why AI adoption has become a concern for customers. Tools are being pushed into daily workflows quickly, while their downsides get little attention. The talk treats AI broadly, covering generative tools, copilots and the models behind them, and asks what professionals give up when they depend on these tools over years.

The middle of the talk walks through four challenges. Misinformation appears because training data is not vetted, and a simple demonstration shows a model giving different answers to a basic question while presenting each one as fact. Over-reliance and skill decay follow, with references to research suggesting that AI assistance speeds people up but can weaken problem-solving, especially for newcomers. Cognitive offloading, where tools take over note-taking and routine tasks, is presented as another way skills fade. The talk then covers opaque training data and accountability, citing a court case in which fabricated legal citations from an AI tool were submitted, and bias in threat prioritisation that can skew SOC and incident response decisions.

The closing section offers three mitigations. The first is an AI-usage policy grounded in the NIST AI Risk Management Framework. The second is reversing the workflow so humans draft first and AI refines the result, which keeps critical thinking active. The third is investing in team building, mentoring and community so foundational skills stay sharp. The speaker ends by asking attendees to check whether their own organisations have AI policies and to reflect on what they are giving up by using these tools.