Keynote
Where
Talks
RVAsec 15
Duration
51 min
A veteran security leader shares field stories and argues the community must rethink how it handles AI-era risk.
Summary
The keynote opens with the speaker's background across decades of security operations, a stint as a defender before moving into vendor work, and a series of anecdotes from the field. Stories about legacy network hardware, a root account that stayed active long after its owner left, and a forgotten server kept running by an abandoned project show how stale systems and credentials quietly pile up risk.
The middle section turns to AI. The speaker describes an internal language model that queried HR records when asked who disliked the user, an email-borne instruction that tricked a commercial assistant into leaking access, and exposed secrets such as API keys and publicly readable cloud storage buckets. He argues that tools are being deployed faster than they are secured, and that the industry is repeating mistakes it made with the internet, Bluetooth and early web products.
The closing part calls for defenders to explain risk in plain language, to build tools that non-engineers can use, and to support one another as careers shift. Practitioners should treat agent credentials, secret handling, patching and regression testing as urgent work, and should keep the human side of the profession in view, including the relationships built at conferences.
Related talks
Talks·47 min
Theresa Lanowitz
Explains how cyber resilience, AI-driven attacks and software supply chain risk shape security priorities for organizations.
Talks·53 min
Joy Curtis
Explains how a project manager's AI deployments in regulated sectors reveal red and green flags around data, outputs and explainability.


