Aguardic logoAguardic
Start Free Trial
Trust Services Criteria

SOC 2 Compliance for AI Systems. Continuous Evidence.

SOC 2 Type II auditors want continuous evidence of security, availability, and confidentiality controls — not point-in-time snapshots. Aguardic maps enforceable policies to Trust Services Criteria CC6–CC8 and generates the audit trail automatically.

Start Free TrialBrowse SOC 2 Pack

14-day free trial · No credit card · Free SOC 2 policy pack

Does This Apply to You?

SOC 2 Is Table Stakes for Enterprise AI

SaaS & Technology

  • B2B SaaS vendors where SOC 2 is table stakes for enterprise sales
  • AI startups building products that process customer data
  • Cloud platforms integrating AI capabilities into existing services

Finance & Healthcare Tech

  • Fintech companies using AI for risk scoring, fraud detection, or credit decisions
  • Healthtech platforms processing sensitive data through AI models
  • Any company where customers require SOC 2 reports before signing

If your customers ask for a SOC 2 report and you use AI in your product, your auditor will expect to see AI-specific controls.

Trust Services Criteria

SOC 2 Controls That Cover Your AI Systems

CC6 — Logical Access

Control who and what can access AI systems and their outputs. Enforce policies on data access, model queries, and output sharing.

CC7 — System Operations

Monitor AI system operations, detect anomalies, and enforce operational policies. Track model performance and availability.

CC8 — Change Management

Track changes to AI models, policies, prompts, and configurations with full audit trail. Document every modification.

Confidentiality

Prevent sensitive data leakage through AI outputs and document sharing. Detect and block confidential information in real time.

Requirements Coverage

SOC 2 Coverage Matrix

No single tool covers every SOC 2 Trust Services Criterion. This is the criterion-to-control reference — what Aguardic enforces, the evidence it produces, and the work your security team still owns.

6Covered
2Partial
3Not Covered
Total: 11
Covered·

CC4.1

Monitoring Activities

Evaluate internal control effectiveness with continuous monitoring, including detection of logging gaps and swallowed errors that undermine audit evidence.

How Aguardic helps

Logging & Monitoring pack detects disabled logging, log-level downgrades, and empty catch blocks that silently swallow errors. Continuous policy evaluation itself is monitoring evidence for auditors.

Evidence produced

Disabled logging detections · swallowed exception records · continuous evaluation logs

What you handle

Run your observability stack (Datadog, Grafana, etc.) and wire alert routing. Aguardic surfaces governance signals, not infrastructure uptime.

Covered·

CC6.1

Logical Access Security

Implement logical access security including credential hygiene for information assets and AI systems that process them.

How Aguardic helps

Access Controls pack detects credential sharing in communications, exposed API keys and tokens, and default credentials. Policy enforcement also gates which users, systems, and AI models reach sensitive data.

Evidence produced

Credential sharing detections · API key exposure alerts · default credential flags · access policy evaluation logs

What you handle

Configure your identity provider, define role mappings, and approve which AI systems operate on sensitive data.

Not Covered·

CC6.2

User Authentication

Authenticate users before granting access to information assets and AI systems.

How Aguardic helps

Aguardic enforces policies after authentication, not the authentication itself. Pair with a dedicated identity provider.

What you handle

Deploy an identity provider with MFA (Okta, Azure AD, etc.) and manage the credential lifecycle for all users and service accounts.

Partial·

CC6.3

Access Authorization

Authorize access based on credentials, system configuration, and business context.

How Aguardic helps

Policies enforce role-based and context-based access to AI systems and their outputs. Does not manage the underlying authorization infrastructure.

Evidence produced

Policy evaluation logs with authorization context

What you handle

Run your authorization infrastructure (IAM, RBAC) and approve the role-based policies Aguardic enforces on top.

Covered·

CC6.7

Transmission & Encryption Controls

Protect information during transmission, movement, and at rest through encryption and approved cryptographic algorithms (CC6.1 / CC6.7 / C1.1).

How Aguardic helps

Data Encryption pack detects unencrypted transmission (FTP, Telnet, plaintext HTTP API), weak algorithms (MD5, SHA-1, DES, TLS 1.0/1.1), plaintext secret storage, and storage configurations with encryption-at-rest disabled.

Evidence produced

Unencrypted transmission detections · weak crypto alerts · plaintext secret flags · encryption-at-rest gap reports

What you handle

Configure TLS on every customer-facing endpoint, rotate certificates and KMS keys, and sign off on the cryptographic standards Aguardic enforces.

Not Covered·

CC7.1

Vulnerability Management

Detect and monitor security vulnerabilities in system components and AI infrastructure.

How Aguardic helps

Aguardic governs AI behavior, not infrastructure security. Pair with dedicated vulnerability scanning.

What you handle

Run vulnerability scanners (Snyk, Dependabot, etc.), patch systems on defined SLAs, and maintain a remediation program.

Covered·

CC7.2

Anomaly Detection

Monitor system components for anomalies indicative of malicious acts or errors.

How Aguardic helps

Continuous policy evaluation across AI outputs detects anomalous behavior, policy violations, and suspicious patterns. Alerts fire in real time.

Evidence produced

Violation alerts · anomaly detection logs · trend reports

What you handle

Tune policy severity thresholds and triage the anomalies Aguardic surfaces during your own security reviews.

Partial·

CC7.3

Security Incident Response

Evaluate detected events and respond to identified security incidents.

How Aguardic helps

Enforcement modes (Block/Warn/Escalate) provide automated first-response for policy violations. Does not replace a full incident-response plan or team.

Evidence produced

Enforcement action logs · escalation records

What you handle

Maintain the full incident-response plan, staff on-call rotations, and run post-incident reviews.

Covered·

CC8.1

Change Management

Authorize, design, develop, test, approve, and implement changes to systems and policies.

How Aguardic helps

Policy versioning tracks every change with timestamp and audit trail. VCS integration enforces code-review policies. All policy changes are logged.

Evidence produced

Policy version history · change-approval logs · VCS evaluation records

What you handle

Define your change-approval workflow (who approves what) and document it in your SOC 2 system description.

Covered·

C1.1

Confidentiality Commitments

Protect confidential information as committed to customers and agreed to contractually.

How Aguardic helps

Content scanning policies detect and block sensitive-data leakage through AI outputs, document sharing, email, and messaging. Enforcement prevents exposure before it happens.

Evidence produced

Data leakage prevention logs · blocked disclosure records

What you handle

Classify what counts as confidential for your org and maintain customer confidentiality commitments in contracts.

Not Covered·

C1.2

Confidential Information Disposal

Dispose of confidential information to meet retention and deletion commitments.

How Aguardic helps

Aguardic prevents leakage but does not manage data retention or disposal.

What you handle

Deploy a data lifecycle management tool, define retention schedules, and execute secure data disposal per your customer commitments.

Browse the SOC 2 Policy Pack

Coverage mappings reflect Aguardic's current product capabilities mapped to SOC 2 Trust Services Criteria (CC6–CC8 and C1). Validate with your auditor for your specific SOC 2 engagement and scope.

SOC 2 with AI questions?

Answer the AI-specific SOC 2 questions with controls Aguardic enforces

Upload it. We draft answers for the AI governance, model access, and AI audit-logging questions — citing what Aguardic enforces. Traditional CC-series controls go to Vanta or Conveyor; we tell you which in the Word doc.

Upload questionnaire

Explore Other Frameworks

HIPAAHealthcareEU AI ActEU RegulationISO 42001AI ManagementColorado AI ActUS StateNIST AI RMFUS FederalAIUC-1AI Agent Security

View all compliance frameworks

Start generating SOC 2 evidence today.

Install the SOC 2 policy pack, connect your AI systems, and get continuous audit evidence from day one.

14-day free trial
No credit card required
Free SOC 2 policy pack
Start Free Trial

Or explore the documentation

SOC 2 Compliance for AI Agents — Continuous Audit Evidence - Aguardic