Dear Readers,
Many of the papers accepted for the 2026 Security and Privacy Symposium are listed with title and authors at the event's web page. They cover a broad range of topics, from "Fast Deterministically Safe Proof-of-Work Consensus" to "Transient Architectural Execution: From Weird Gates to Weird Programs", and "Searching for a Farang: Collective Security among Women in Pattaya, Thailand", and "MusicShield: Protection for Musicians in the Era of Generative AI". The conference runs from May 18-21 in San Francisco.
A few of the papers address topics in use of AI, but we suspect that agentic AI is moving faster than even the most agile peer review systems. Unlike quantum computing, AI agents are moving at breakneck speed. These marvelous code generation systems will soon dominate the battle of good and evil on the computer security front. We predict that most research papers about security from here on will be about autonomous AI agents. Let's hope that the defensive side can move faster than the offensive side.
One of the dangers of these systems is the immense amount of computing power being put in place to support them. An evil collection of AI agents can rip through widespread password guessing, zero day probing, exfiltration, and other malicious tasks with unprecedented speed. It is not necessary to for an individual to acquire resources for these tasks, the resources are part of the AI systems. Of course, vendors and governments will try to prevent use of the systems for evil purposes, but that will simply be part of a game in with AI is used to design evasions of AI policing.
New Wars, Old Songs Made New