If the earliest fully autonomous AI agents end up being malicious, we will need some sort of identity verification/attribution standard @KeycardLabs is the only one I know focused 100% on building a solution for this, but others like @ai_pensar, @zenitysec, and @DuneSecurity are exploring it as part of a wider suite of agent security products It's a super nascent niche of the agentic ecosystem that's been fun to follow