AI risk and security certification vs. outcome verification

  • Comparison
  • Published
  • Reviewed
  • By

Certifications and attestations check the controls around an AI agent: whether it resists jailbreaks and data leakage (AIUC-1), whether the service's security and privacy controls work (SOC 2), whether the organization manages AI under a defined system (ISO/IEC 42001). Outcome verification checks something none of them claims to measure: whether the outcomes the agent reported for real customers held up.

What each says about itself#

  • AIUC-1 is a certification for AI agent security, safety and reliability across six domains (data and privacy, security, safety, reliability, accountability, society). It uses evaluations of thousands of real-world scenarios and red-teaming, an accredited auditor reviews policy, operational and technical controls, the certificate is valid for a year, and agents are retested quarterly (AIUC-1, updated 15 October 2026).
  • AICPA and AIUC announced on 19 October 2026 a single CPA-led engagement combining a SOC 2 report with an AIUC-1 certification (AICPA and CIMA).
  • SOC 2 is an examination of a service organization's controls relevant to security, availability, processing integrity, confidentiality or privacy, by a licensed CPA (AICPA and CIMA).
  • ISO/IEC 42001 specifies requirements for an AI management system within an organization (ISO).

Side by side#

Certification or attestationOutcome verification (Attify)
MeasuresControls and behaviour under testsOutcomes in production
AnswersIs it safe and well run?Did its claimed outcomes happen, stick, and need no human?
EvidencePolicies, records, adversarial scenarios, auditor testingThe customer's own help desk, CRM, calendar or booking system
CadenceAnnual, with quarterly retests (AIUC-1)Daily rules tier, judged scores every 30 days, live status
OutputA certificate and an audit reportAn audited rate with an interval and a signed attestation; a public profile and badge once the vendor publishes

Using both#

Where procurement requires a certification before connecting any customer data, that is the right order: check that the agent is safe to connect, then check what it achieves once connected. The two are complements; neither substitutes for the other.

Attify's own security posture is described on /security.

Questions

If a vendor is AIUC-1 certified, do I still need to verify outcomes?

Yes, if outcomes matter to you. AIUC-1 tests safety, security and reliability behaviour against adversarial scenarios. It does not say it measures whether issues were resolved or appointments kept.

Is Attify a certification?

No. Attify measures outcomes under a published standard per job type; the scores a vendor publishes carry a live status and signed attestations. It does not certify controls, security or management systems.

Sources

  1. AIUC. AIUC-1. Updated 15 Oct 2026. Read 8 Nov 2026.
  2. AICPA and CIMA. Auditors meet urgent demand for AI assurance with new offering from AICPA and AIUC. Published 19 Oct 2026. Read 8 Nov 2026.
  3. AICPA and CIMA. SOC suite of services. Read 8 Nov 2026.
  4. ISO. ISO/IEC 42001:2023 Artificial intelligence: Management system. Read 8 Nov 2026.