The Automation of Exploits: Will Eatherton on AI-Driven Security and the End of Obscurity
summary
Claims that LLMs now chain firmware, board schematics, FPGA designs and embedded code into reproducible attacks and Cisco experiments show concrete results: an automated assembly line that identifies a vulnerability, generates a config, flashes firmware or FPGA bitstreams, and emits test traffic to validate an exploit in minutes instead of days. The team refactors legacy C to Rust at scale, sees hundreds of exploits churned by automation, and recommends moving to quarterly patch cadence.
AI-written summary · about
chapters (8)
- 0:00Introduction at Hardware IO 2026
- 0:23Attackers evolving with AI tools
- 1:46Software and hardware convergence
- 3:52Domain-specific AI models needed
- 5:57Open-source hardware future
- 7:43Hardware complexity and security
- 9:50Faster software updates necessity
- 11:13Recommended lab tools for engineers
related talks
39mThe AI Arms Race in Vulnerability Management, Who’s Winning?
29mBuilding a Hackbot: How Ethiack's Hackian Finds Real Vulnerabilities
29mThe vulnerability apocalypse: AI broke the security playbook
57mAI as the New Hacker Frontier
40mAI As The New Hacker Frontier
44mTL;DR: Applying AI to Security
49mAutonomous Hacking Systems - Future Risk or FUD?
20mKeynote: Outsmarting IoT Defense: The Hacker's Perspective