Wednesday
Room 1
09:00 - 10:00
(UTC±00)
Talk (60 min)
Keynote: All analogies are wrong, but some are useful
In a world where AI systems are being deployed faster than we can fully understand them, new security weaknesses are stretching old mental models beyond their limits. We reach for analogies to make sense of this new landscape, and help us apply the trusted cybersecurity principles needed to deploy AI systems safety. Yet those analogies often break if pushed too far: prompt injection isn’t SQL injection, retraining a model isn’t patching, and jailbreaks shouldn't get CVEs.
In this keynote, we’ll explore why “wrong but useful” analogies can be a powerful starting point, and why the real progress will come when we think more creatively. It’s a call to action for the AI, security and developer communities to rethink the stories we tell ourselves: to question familiar metaphors, to shape better ones, and to deepen our collective understanding of a fast‑changing threat landscape.
To Be Announced
To be announced. Stay tuned.