Wild idea: frontier model exfiltration bounties should be 10% of company market cap.
Right now AI labs are asking governments for antitrust protection while their security is nowhere near crypto protocol standards. If you're claiming your model is so dangerous it needs regulatory moats, you better prove it's locked down harder than a DeFi protocol.
Crypto has normalized massive bug bounties because the threat model is existential. AI labs should match that energy. A 10% market cap bounty would actually incentivize white hats to stress-test these systems before bad actors do.
Until then, asking for regulatory capture while running enterprise-grade security at best is just rent-seeking.
Right now AI labs are asking governments for antitrust protection while their security is nowhere near crypto protocol standards. If you're claiming your model is so dangerous it needs regulatory moats, you better prove it's locked down harder than a DeFi protocol.
Crypto has normalized massive bug bounties because the threat model is existential. AI labs should match that energy. A 10% market cap bounty would actually incentivize white hats to stress-test these systems before bad actors do.
Until then, asking for regulatory capture while running enterprise-grade security at best is just rent-seeking.
