Undated
AIUC has published crosswalks mapping its standard against about ten other frameworks, publicly available on its website.
“you'll find that we've done crosswalks. I think it's about ten different frameworks now that they're transparently available”
Emil Lassen · AIUC-1: Building trust in AI agents, 0:16:25
Undated
Six of the 40 mandatory requirements in AIUC-1 concern red teaming, meaning testing whether technical controls hold up under pressure.
“six of the 40 mandatory requirements in AIUC-1 have to do with red teaming, actually testing that these technical controls and hold up under pressure.”
Emil Lassen · AIUC-1: Building trust in AI agents, 0:21:31
Undated
Lassen says AIUC-1's most recent quarterly update prioritized risks from MCP, a way for agents to connect to tools and exchange information.
“Last quarter, it was MCP risk, for example, which has really come up as agents start not just operating in isolation, but exchanging information.”
Emil Lassen · AIUC-1: Building trust in AI agents, 0:22:34
Undated
Companies get roughly one to four weeks between red-team rounds to mitigate findings, depending on how serious they are.
“depending on the magnitude of the findings, your team will have between, say, one and four weeks to mitigate these things”
Emil Lassen · AIUC-1: Building trust in AI agents, 0:28:42
Undated
The final AIUC-1 audit report is between 60 and 100 pages long and is meant to help unblock enterprise deals.
“It's between 60 and 100 pages long, and it's an asset you can really unblock those enterprise deals with.”
Emil Lassen · AIUC-1: Building trust in AI agents, 0:29:12
Undated
AIUC re-runs its red-team tests every quarter, and passing them is a requirement to keep the AIUC-1 certification.
“And we do that every single quarter, and that's a requirement to maintain the certification.”
Emil Lassen · AIUC-1: Building trust in AI agents, 0:29:37
Undated
Under AIUC-1 grading, a company cannot pass if red teaming finds any P0 or P1 vulnerabilities, the two most severe levels.
“you cannot pass AC one if you have any P zero or P one vulnerabilities identified, you have to mitigate those.”
Emil Lassen · AIUC-1: Building trust in AI agents, 0:31:00
Undated
Lassen says White Circle's monitoring and filtering tools have helped companies meet many safety requirements in AIUC-1, though AIUC does not work with White Circle.
“Their monitoring and filtering work is really good and has helped companies meet a lot of the safety requirements in our standard.”
Emil Lassen · AIUC-1: Building trust in AI agents, 0:41:24