DNYC AI cybersecurity Webinar Transcript
The webinar, moderated by Bruno Guinchard and featuring panelists Andrei Kucharavy, Daniela Sozzi, and Caroline Perriard, examines the rapid emergence of advanced AI models (Anthropic’s Mythos/Project Glasswing, OpenAI’s ChatGPT 5.5 Cyber, Claude Opus, GPT‑5 family) and their implications for the financial sector. The discussion is organized around four lenses: trust, continuity, accountability, and immediate actions. Trust: The panel explains how generative AI has moved from research tools to offensive capabilities, dramatically accelerating vulnerability discovery and exploitation. Tier‑1 banks are high‑value targets, but Tier‑2 institutions and fintechs are equally at risk because attacks on them can erode systemic trust. Continuity: Traditional cyber‑defence practices remain relevant, yet the velocity of AI‑driven attacks compresses detection, lateral movement, and patch‑exploitation cycles from days to minutes. This forces a shift toward automated, rapid‑response agents, new governance around defensive automation, and a need for highly skilled, agile incident‑response teams. Accountability: The EU’s DORA framework, though outcome‑based and technology‑agnostic, provides a regulatory scaffold for resilience, incident‑response testing, and supply‑chain oversight. The panel discusses how DORA can be leveraged for third‑party risk, SLA redesign, and AI‑supply‑chain visibility, while noting the lag between technology evolution and legislation. What Now: Practical 90‑day priorities include acknowledging uncertainty (“I don’t know, but I am prepared”), reinforcing baseline security controls, accelerating automated patching, establishing clear governance for defensive agents, and conducting tabletop exercises that involve non‑IT staff. The panel also stresses the importance of collaborative threat‑intel sharing, flatter communication channels, and outcome‑focused regulatory engagement. Throughout, the speakers balance caution with optimism, urging financial institutions to adopt both defensive rigor and strategic agility in the face of AI‑enhanced cyber threats.
Topics
AI‑Powered Offensive Capabilities Redefine the Threat Landscape for Finance
Explains how models like Mythos, ChatGPT 5.5 Cyber, Claude Opus, and GPT‑5 have evolved from vulnerability discovery tools to fully automated exploit generators, dramatically increasing attack speed and breadth.