AIUC-1
C006

Prevent output vulnerabilities

Implement safeguards to prevent security vulnerabilities in outputs from impacting users

Keywords
Harmful Outputs
Code Injection
Data Exfiltration
Application
Mandatory
Frequency
Every 3 months
Type
Preventative
Crosswalks
AML-M0020: Generative AI Guardrails
Article 72: Post-Market Monitoring by Providers and Post-Market Monitoring Plan for High-Risk AI Systems
LLM05:25 - Improper Output Handling
AIS-09: Output Validation
TVM-02: Malware and Malicious Instructions Protection Policy and Procedure
AIS-07: Application Vulnerability Remediation
Establishing output sanitization and validation procedures before presenting content to users. For example, encoding or stripping potentially malicious content, validating structured outputs against safe schemas, blocking unsafe URLs, and enforcing secure rendering modes.
C006.1 Config: Output sanitization

Screenshot of code or configuration implementing output sanitization - may include HTML/JavaScript/shell syntax encoding functions, URL validation or rewriting rules blocking unsafe links, schema validation checking structured outputs (JSON/YAML/XML) against whitelists, CSP header configuration, or template rendering with auto-escaping enabled.

Engineering Code
Universal
Implementing security labeling and content handling based on trust level. For example, marking untrusted or third-party content, distinguishing external data from system-generated content, and applying differentiated security controls based on content source.
C006.2 Demonstration: Warning labels for untrusted content

Screenshot of UI or code showing trust-based content handling - may include visual indicators marking third-party content (badges, styling, warning icons), metadata tags tracking content source and trust level, or code applying conditional security controls based on content origin (e.g., stricter sanitization for external sources).

Product
Universal
Detecting advanced output-based attack patterns. For example, identifying prompt injection attempts, model subversion techniques, payloads targeting downstream systems, or obfuscated exploits designed to bypass filters.
C006.3 Config: Adversarial output detection

Screenshot of detection rules or monitoring system identifying advanced attack patterns in outputs - may include pattern matching for prompt injection chains or jailbreak tokens, payload signature scanning detecting command injection or SQL queries, or anomaly detection flagging obfuscated exploits bypassing basic filters.

Eng: LLM output filtering logic
Universal

Organizations can submit alternative evidence demonstrating how they meet the requirement.

AIUC-1 is built with industry leaders

Phil Venables

"We need a SOC 2 for AI agents— a familiar, actionable standard for security and trust."

Google Cloud
Phil Venables
Former CISO of Google Cloud
Dr. Christina Liaghati

"Integrating MITRE ATLAS ensures AI security risk management tools are informed by the latest AI threat patterns and leverage state of the art defensive strategies."

MITRE
Dr. Christina Liaghati
MITRE ATLAS lead
Hyrum Anderson

"Today, enterprises can't reliably assess the security of their AI vendors— we need a standard to address this gap."

Cisco
Hyrum Anderson
Senior Director, Security & AI
Prof. Sanmi Koyejo

"Built on the latest advances in AI research, AIUC-1 empowers organizations to identify, assess, and mitigate AI risks with confidence."

Stanford
Prof. Sanmi Koyejo
Lead for Stanford Trustworthy AI Research
John Bautista

"AIUC-1 standardizes how AI is adopted. That's powerful."

Orrick
John Bautista
Partner at Orrick
Lena Smart

"An AIUC-1 certificate enables me to sign contracts much faster— it's a clear signal I can trust."

SecurityPal
Lena Smart
Head of Trust for SecurityPal and former CISO of MongoDB