Who this is for: Cloud service providers offering AI capabilities to federal agencies, GRC teams managing FedRAMP packages, C3PAOs performing assessments, ISSMs and ISSOs overseeing cloud deployments, and compliance engineers building continuous authorization pipelines.

ACTIVE TRANSITION: FedRAMP 20x is live. The Consolidated Rules for 2026 (CR26) define requirements for cloud providers, assessors, and agencies. Transition deadlines extend through 2028. Existing FedRAMP authorized systems must transition to the 20x framework during their next annual assessment.

CRITICAL ASSESSOR NOTICE: SWT3 witness anchors record that a governance event occurred and capture its computational factors. They do not replace assessor judgment. The assessor must independently verify that the substance of each control implementation meets FedRAMP 20x requirements. Anchors provide the evidence trail -- the assessor determines whether that evidence is sufficient.

Contents

1. Quick Reference 2. What Changed: Legacy vs 20x 3. Key Security Indicators (KSIs) 4. AI-Prioritized Authorization 5. Obligation-to-Procedure Mapping 6. Detailed Procedure Cards 7. SDK Integration Examples 8. Related Guides and References

1. Quick Reference

FieldDetail
ProgramFedRAMP 20x -- Consolidated Rules for 2026 (CR26)
AuthorityFedRAMP Program Management Office (GSA)
EffectiveCR26 published June 2026; transition deadlines through 2028
Key Terminology"FedRAMP Certified" replaces "FedRAMP Authorized"
Authorization TimelineLow/Moderate: ~3 months (down from 18+ months via automation + KSIs)
AI PriorityPhase 2 submissions included AI-prioritized offerings and critical-need GRC tools
ConMon ReplacementVulnerability Detection and Response (VDR) standard -- accounts for rapid-turnaround AI threats
New ConceptKey Security Indicators (KSIs) -- automated, measurable compliance signals
System ClassificationClass A (low-risk SaaS), Class B (moderate), Class C (high/sensitive), Class D (critical infrastructure)
SWT3 Procedures10 procedures mapped across 8 namespaces

2. What Changed: Legacy vs 20x

DimensionLegacy FedRAMPFedRAMP 20x (CR26)
Status Term"FedRAMP Authorized""FedRAMP Certified"
Authorization Timeline12-18+ months~3 months (Low/Mod pilot)
Continuous MonitoringConMon (monthly scans, annual assessments)VDR standard (rapid detection, AI threat-aware)
Assessment Model3PAO annual assessmentAutomation-first + KSI-driven + periodic 3PAO
Evidence FormatStatic SSP documents, spreadsheetsMachine-readable KSIs, automated evidence pipelines
AI ConsiderationNo AI-specific guidanceAI-prioritized authorization path, AI threat categories in VDR
System ClassificationLow / Moderate / HighClass A through D (risk-based, broader spectrum)

Transition path: Existing "FedRAMP Authorized" systems retain their status. During the next annual assessment, they begin the transition to FedRAMP 20x requirements. No cliff-edge -- the transition is phased through 2028. New authorizations follow the 20x path exclusively.

3. Key Security Indicators (KSIs)

KSIs are the core innovation in FedRAMP 20x. They replace static document reviews with automated, measurable compliance signals that can be continuously verified. KSIs are grouped by domain:

SWT3 witness anchors map naturally to KSIs. Each anchor is a machine-readable, cryptographically verifiable compliance signal -- exactly the type of evidence KSIs are designed to consume.

4. AI-Prioritized Authorization

FedRAMP 20x Phase 2 explicitly prioritized AI-focused cloud service offerings alongside trust centers and GRC platforms. This signals that the PMO recognizes AI services require authorization pathways that match the pace of AI deployment in federal agencies.

For AI service providers, the 20x path offers several advantages:

5. Obligation-to-Procedure Mapping

FedRAMP 20x ObligationProcedureDescriptionKSI Domain
AI governance structure AI-GOV.1 Governance event witnessing for authorization package Governance
VDR continuous detection AI-DRIFT.1 Drift and vulnerability detection for AI components Vulnerability
Authorization audit trail AI-AUDIT.1 Audit integrity witnessing for certification evidence Audit
AI lifecycle documentation AI-LCM.1 Lifecycle event witnessing for authorization package Governance
Supply chain risk AI-SUPPLY.1 Supply chain provenance witnessing Supply Chain
AI component SBOM AI-SBOM.1 Software bill of materials for AI components Supply Chain
AI security baseline AI-SEC.1 Security control witnessing per KSI baseline Security
Performance monitoring AI-PERF.1 Performance metric witnessing against KSI thresholds Performance
System robustness AI-ROBUST.1 Robustness and failure recovery witnessing Resilience
Infrastructure boundary AI-INF.1 Infrastructure witnessing for system boundary documentation Infrastructure

6. Detailed Procedure Cards

AI-GOV.1

AI Governance Witnessing

FedRAMP 20x requires: Cloud service providers must document and maintain an AI governance structure as part of the authorization package. The governance KSI verifies that governance is not just documented but actively maintained -- meeting schedules, policy reviews, and decision records must be demonstrably current.

How SWT3 addresses it: witnessGovernance() mints an anchor recording the governance event type (policy review, board meeting, risk acceptance), participants, decision outcome, and timestamp. Each anchor produces a KSI-compatible evidence artifact that can be fed directly into automated compliance verification pipelines. The anchor chain proves governance is continuous, not point-in-time.

Assessor Tip

Query AI-GOV.1 anchors for the assessment period. Verify governance meeting cadence matches the organization's stated schedule. For 20x, evidence of continuous governance activity is more valuable than a single annual review document.

AI-DRIFT.1

Vulnerability Detection and Response

FedRAMP 20x requires: The VDR standard replaces legacy ConMon. It requires rapid detection and response to vulnerabilities, including AI-specific threat categories: model poisoning, prompt injection, training data contamination, and adversarial inputs. Response timelines are compressed compared to legacy ConMon, reflecting the speed at which AI vulnerabilities can be exploited.

How SWT3 addresses it: witnessDrift() mints an anchor recording the drift type (model drift, data drift, security drift), detection method, severity, and response action. For VDR compliance, the anchor chain proves that detection is continuous and response is timely. Factor A captures the detection signal type. Factor B records the response latency in milliseconds.

Assessor Tip

AI-DRIFT.1 anchors should demonstrate continuous monitoring, not periodic scans. Under VDR, look for detection-to-response latency in Factor B. AI-specific threats (model poisoning, prompt injection) should appear in the anchor observations alongside traditional CVE-based vulnerabilities.

AI-SUPPLY.1

Supply Chain Provenance

FedRAMP 20x requires: AI cloud services must document all third-party AI components including model providers, training data sources, fine-tuning pipelines, and inference dependencies. The supply chain KSI verifies that this inventory is complete and that risk assessments are current for each component.

How SWT3 addresses it: witnessSupplyChain() mints an anchor recording the component inventory hash, provider attestation status, and last risk assessment date. The anchor chain proves that supply chain visibility is maintained across model updates, provider changes, and dependency upgrades.

Assessor Tip

Verify that AI-SUPPLY.1 anchors cover every AI component in the system boundary diagram. Cross-reference with AI-SBOM.1 anchors to confirm the SBOM and supply chain inventory are consistent. Any component in the SBOM without a corresponding supply chain risk assessment is a finding.

AI-SEC.1

AI Security Baseline

FedRAMP 20x requires: AI-specific security controls must be implemented per the KSI baseline for the system's classification (Class A through D). This includes prompt injection defenses, model access controls, output filtering, and inference pipeline integrity. The security KSI provides continuous verification that these controls are active and effective.

How SWT3 addresses it: witnessSecurity() mints an anchor recording the security control category, implementation status, test results, and last verification date. Each anchor maps directly to a security KSI measurement. The anchor chain demonstrates that AI security controls are not just implemented but continuously verified.

Assessor Tip

For Class B and above, verify that AI-SEC.1 anchors cover all AI-specific threat categories defined in the VDR standard. Automated security KSI signals should align with anchor evidence. Any gap between KSI reporting and anchor evidence warrants investigation.

7. SDK Integration Examples

Python

pip install swt3-ai

from swt3_ai import Witness
witness = Witness(api_key="axm_live_xxx", strict=True)

# Witness AI governance event for FedRAMP 20x KSI
witness.witness_governance(
    event_type="policy_review",
    decision="approved",
    participants=["ciso", "ai_governance_board"]
)

# Witness VDR detection event
witness.witness_drift(
    drift_type="security_drift",
    detection_method="automated_scan",
    severity="high",
    response_action="patch_deployed"
)

# Witness supply chain component update
witness.witness_supply_chain(
    component="openai-gpt4o",
    risk_assessment="current",
    last_review="2026-08-27"
)

TypeScript

npm install @tenova/swt3-ai

import { Witness } from '@tenova/swt3-ai';
const witness = new Witness({ apiKey: 'axm_live_xxx', strict: true });

// Witness AI security control verification for KSI
await witness.witnessSecurity({
    controlCategory: 'prompt_injection_defense',
    implementationStatus: 'active',
    testResult: 'pass',
    lastVerified: '2026-08-27'
});

await witness.flush();

8. Related Guides and References