Defending Multisigs from Offensive AI: Introducing onchain security markets

Offensive AI is increasing the speed and scale of onchain attacks. Yet transaction security still relies on opaque, centralized checkers and UI warnings. Users cannot inspect how decisions are made, compare providers, or hold them accountable. A warning also does not prevent execution.

Onchain security markets offer a different model. Independent human and AI defenders evaluate defined risks, validators attest to the results, and smart-account guards enforce the outcome before execution. This opens checks to competing specialists and makes outputs transparent, verifiable, and accountable.

Using Safenet as a case study, the talk examines common attack patterns such as address poisoning, malicious approvals, and unauthorized account changes. Attendees will learn how onchain security markets can turn opaque warnings into enforceable account policy, and how the design handles disagreement, downtime, incorrect blocking, and recovery.

Speakers: Rahul Rumalla

Track: Security