
Tue Nov 17 · 4:30pm
Provably Secure: Math Against Superintelligent Hackers
A panel from MIT’s Social and Ethical Responsibilities of Computing asks whether formal verification, mathematical proof about how programs behave, could guarantee AI agent containment and find every security bug of certain kinds as LLMs get better at finding vulnerabilities. Adam Chlipala of MIT moderates Byron Cook, vice president and distinguished scientist at Amazon Web Services; Noah Evans of Draper; and MIT’s Armando Solar-Lezama.