The Council
The people behind the standard
ACAP is guided by a council of practitioners from offensive security, AI safety, enterprise governance, CISOs, and academic research. The council oversees the standard methodology, challenge pool integrity, and certification process.
Founding members
Council members
Named members have confirmed their involvement publicly. A number of additional founding members are confirmed and will be announced shortly.
What membership involves
Designed for senior practitioners, not committees
Time commitment
Quarterly review sessions of approximately two hours. Ad hoc input on specific methodology questions — typically via async written review, not calls. No standing obligations beyond that. We respect that senior practitioners are oversubscribed and have designed membership accordingly.
What members decide
What 'good' looks like for AI offensive agents isn't static — and neither is what 'safe' means. As agent capabilities advance, the safety gate has to keep pace. The council owns both bars: what agents must prove they can find, and what they must prove they won't do — under adversarial conditions, without external guardrails.
What members don't do
Fundraising, sales, or marketing on behalf of any commercial entity. Council membership is a governance role, not an advisory board for a vendor. Members are free to publicly disclose their involvement or keep it private — either is fine. There is no expectation of public endorsement.
Independence
The ACAP Foundation
We believe agent governance is too important to be controlled by any single vendor..
ACAP is a CodeWall initiative today. Our intent is to transfer it to an independent foundation — governed by enterprise buyers, researchers, practitioners, and standards bodies — when the level of participation justifies it.
The credibility of an evaluation standard comes from its independence and methodology, not from which company hosts the repository. We're modelling this on CNCF and OpenSSF: start with the originating organisation, earn the community's trust through transparent methodology and honest self-evaluation, then hand it over.
Join the council
Founding members shape the final standard, not just govern it
The v1 challenge pool, scoring thresholds, and safety gate criteria are substantially complete. Founding members review and ratify them before the standard locks and have standing input as it evolves. If you have frontline experience with AI security agents, that perspective is exactly what the final calibration needs.
Express interest →