PricingDemo
Log InGet API Key
Integrity Infrastructure — Real-Time

Continuous Attestation

Every control state change produces a cryptographic receipt. Post-quantum signed. Independently verifiable. Replay-grade. Not a dashboard metric. Not a periodic scan. A deterministic, tamper-evident record of exactly what happened, when it happened, and under whose authority it happened.

42 µs
Per Attestation
74 B
H33-74 Receipt
3
PQ Signature Families
Replay Horizon
Definition

What Continuous Attestation Means

Continuous attestation is the practice of producing a cryptographic receipt for every control state change in an organization. Not every hour. Not every day. Every change.

When an administrator modifies a multi-factor authentication policy, a signed attestation record is produced in the same transaction. When an encryption key is rotated, the rotation event is attested before the new key becomes active. When an AI agent's scope is expanded or restricted, the scope modification generates a receipt containing the previous authority boundary, the new authority boundary, the authorizing principal, and a post-quantum signature binding all of those fields together.

This is not monitoring. Monitoring observes. Attestation proves. A monitoring system can tell you that a control changed. An attestation record proves what the control was before the change, what it became after the change, who authorized the change, what policy governed the change, and whether the change was compliant at the instant it occurred.

The distinction matters because monitoring data is retrospective and mutable. An attacker who compromises a SIEM can alter log entries. An insider who modifies access policies can suppress the change notification. But an attestation record that has been signed with three independent post-quantum signature families and anchored to a governance chain cannot be retroactively modified without invalidating every subsequent record in the chain.

The Attestation Receipt

Every attestation produces an H33-74 receipt: 74 bytes that encode the complete cryptographic proof of a state transition. The receipt contains a hash commitment to the prior state, a hash commitment to the new state, the authority chain that authorized the transition, and a post-quantum signature that binds these fields together. The receipt is deterministic: given the same inputs, any independent verifier will produce the same receipt. This is what makes governance replay possible.

Attestation receipts are not stored in a proprietary database. They are anchored to an append-only governance chain that any third party can verify independently. The chain structure ensures that tampering with any individual record invalidates every subsequent record, making selective modification detectable without trusting H33 infrastructure.

The Problem

Why Periodic Audits Fail

The cybersecurity industry operates on a model of periodic validation. Annual penetration tests. Quarterly vulnerability scans. Monthly access reviews. SOC 2 Type II audits that cover a defined observation period. The assumption is that if controls are verified at regular intervals, the organization is secure between those intervals.

This assumption is wrong, and the data proves it. The average time from initial compromise to detection is 204 days. The average time between SOC 2 observation windows is 365 days. The gap between what periodic audits verify and what actually happens in production is measured in months of unmonitored drift.

Drift Is the Vulnerability

Control drift is not a theoretical risk. It is the primary mechanism by which security failures occur in mature organizations. A well-configured firewall that passes an annual audit can be misconfigured the following week by an engineer responding to a production incident. An MFA policy that is verified during a compliance review can be silently downgraded by an administrator who receives too many helpdesk tickets. An encryption-at-rest configuration that meets regulatory requirements during assessment can be disabled during a database migration and never re-enabled.

None of these scenarios involve malice. They involve the ordinary pressure of operations acting on controls that are only verified periodically. The longer the interval between verifications, the more time drift has to compound.

The Evidence Gap

When a breach occurs, the first question from insurers, regulators, and legal counsel is the same: were controls in place at the time of the incident? Periodic audits cannot answer this question. An audit that passed six months before the breach proves nothing about control state at the moment of compromise. Log files may or may not exist. SIEM data may have been rotated or corrupted. The organization is left arguing from inference rather than evidence.

Continuous attestation eliminates this gap. Every control state change is recorded with cryptographic precision. The governance chain contains a complete, tamper-evident history of every policy modification, every key rotation, every access grant, every encryption state transition. When the breach investigation begins, the evidence already exists.

Architecture

How Continuous Attestation Works

Four stages. Every control change. No exceptions.

01

Control Change Detection

H33 hooks into the control plane of your infrastructure: identity providers, key management systems, policy engines, encryption configurations, AI agent frameworks. When any governed control changes state, the change event is captured with the prior state, the new state, the authorizing principal, and the governing policy version.

02

Hash Commitment

The state transition is serialized using deterministic canonical encoding. A SHA3-256 hash of the prior state and a SHA3-256 hash of the new state are computed. These hashes, together with the authority chain, policy reference, and timestamp, form the attestation payload. The encoding is deterministic: identical inputs always produce identical byte sequences, which is the foundation of replay-grade verification.

03

Post-Quantum Signature

The attestation payload is signed using three independent post-quantum signature families backed by three independent hardness assumptions. A forged attestation would require simultaneously breaking lattice problems, NTRU structures, and stateless hash functions. The resulting signature bundle is compressed into an H33-74 receipt: 74 bytes of post-quantum attested proof.

04

Governance Chain Append

The signed attestation is appended to an ordered governance chain. Each record contains the hash of the previous record, creating a tamper-evident linked structure. Any modification to a historical record invalidates every subsequent record in the chain. The chain is independently verifiable: no trust in H33 infrastructure is required to validate chain integrity.

{
  "attestation_id": "att_8f3c2a1b",
  "control_type": "mfa_policy",
  "prior_state_hash": "sha3_a7f1c3...",
  "new_state_hash": "sha3_9e2b4d...",
  "authority": "admin@org.example",
  "policy_ref": "POL-2026-047-v3",
  "timestamp": "2026-05-18T14:32:07.831Z",
  "chain_position": 847291,
  "prior_record_hash": "sha3_b4e8f2...",
  "h33_74_receipt": "0x4a8f...2c1b",
  "signature_families": ["ML-DSA", "FALCON", "SLH-DSA"],
  "replay_integrity": "deterministic"
}
Coverage

What Gets Attested

Continuous attestation is only as valuable as the breadth of controls it covers. H33 attests the complete surface area of organizational governance, not a curated subset.

Control DomainAttested EventsAttestation Trigger
Multi-Factor AuthenticationPolicy changes, method additions/removals, enforcement level modifications, bypass grantsPolicy write
Encryption ConfigurationAlgorithm changes, key length modifications, mode transitions, at-rest/in-transit state changesConfig commit
Key LifecycleGeneration, rotation, revocation, expiration, escrow, recoveryKey operation
Access PoliciesRole grants, permission changes, group modifications, conditional access rulesIAM mutation
AI Agent ScopeAuthority boundary changes, tool access modifications, model deployment approvals, scope expansionsAgent config change
Compliance PostureFramework alignment changes, control mapping updates, evidence collection statePosture transition
Data ClassificationClassification level changes, handling requirement modifications, retention policy updatesClassification event
Governance DecisionsPolicy approvals, exception grants, risk acceptances, remediation completionsDecision record

Every attested event includes the prior state, the new state, the authorizing principal, the governing policy version, and a post-quantum signature. No event is attested without the full context of why it happened.

Analysis

Periodic Attestation vs Continuous Attestation

The difference between periodic and continuous attestation is not incremental. It is structural.

DimensionPeriodic AttestationContinuous Attestation (H33)
Verification FrequencyQuarterly / AnnualEvery state change
Drift DetectionDays to monthsZero latency
Evidence TypeScreenshots, reports, questionnairesCryptographic receipts
Tamper ResistanceNone (editable documents)PQ-signed governance chain
Independent VerificationRequires trust in auditorAny party, any time
Quantum ResistanceNot addressedThree hardness assumptions
Historical ReconstructionIncomplete / best-effortDeterministic replay to any timestamp
Breach InvestigationForensic reconstructionReplay the governance chain
Insurance EvidenceSelf-reportedMachine-verifiable attestation stream
Compliance ProofPoint-in-time snapshotContinuous evidence of control state
Technical Architecture

Attestation Pipeline Internals

Signature Composition

H33 attestation signatures are not single-algorithm signatures. Each attestation payload is signed independently by three post-quantum signature families: ML-DSA (lattice-based), FALCON (NTRU-lattice-based), and SLH-DSA (stateless hash-based). These three families are backed by three independent mathematical hardness assumptions. Breaking one family does not weaken the other two.

The three signatures are composed into an H33-74 receipt through a distillation process that reduces the combined signature material to 74 bytes while preserving independent verifiability. The distillation is one-way: the 74-byte receipt can be verified against the original attestation payload, but the original signatures cannot be reconstructed from the receipt alone. This property is critical for storage efficiency without sacrificing verification capability.

Chain Structure

The governance chain is an append-only sequence of attestation records. Each record contains the SHA3-256 hash of the previous record, creating a linked structure where any modification to a historical record propagates forward through every subsequent hash. An independent verifier can validate the entire chain by recomputing hashes from the genesis record forward.

Chain integrity does not depend on a central authority. The chain structure itself provides tamper evidence. If H33 infrastructure were to be compromised and an attacker attempted to modify a historical attestation, every subsequent record's hash would fail verification. The independent verifier detects the modification without needing to trust any party, including H33.

Deterministic Encoding

Replay-grade attestation requires deterministic encoding. The same state transition, attested at the same time, by the same authority, must always produce identical byte sequences. H33 uses canonical encoding rules that specify field ordering, byte-level serialization, and domain separators for every attestation type. These encoding rules are frozen in the HATS protocol and cannot change without a breaking version increment.

Deterministic encoding is what makes governance replay possible. Because encoding is deterministic, any independent implementation that follows the HATS encoding specification will produce identical attestation records for identical inputs. This means governance replay is not limited to H33 infrastructure. Any conformant implementation can replay the governance chain and arrive at identical control state reconstructions.

Latency and Throughput

Attestation latency is dominated by the post-quantum signature computation. On H33 production infrastructure (Graviton4, ARM64), a single attestation completes in 42 microseconds. Batch attestation processes 32 state transitions in a single cryptographic operation, producing individual receipts for each transition while amortizing signature cost across the batch. The attestation pipeline is built in Rust with no JavaScript in the cryptographic hot path.

Underwriting

Continuous Evidence for Insurance Underwriting

Cyber insurance underwriting is broken. Underwriters evaluate risk based on application questionnaires and annual assessments. Policyholders self-report their security posture. There is no mechanism for the underwriter to independently verify that controls described in the application are actually in place, let alone that they remain in place throughout the policy period.

Continuous attestation changes the economics of underwriting. Instead of relying on self-reported snapshots, underwriters receive a cryptographically signed evidence stream that proves control state at every moment of the policy period. The evidence is not generated by the policyholder's security team. It is generated by the cryptographic infrastructure that governs the controls themselves.

Claims Verification

When a claim is filed, the insurer can replay the governance chain to reconstruct the exact control state at the time of the incident. Were MFA policies enforced? Was encryption at rest enabled? Were access controls properly configured? The answers are not opinions or interpretations. They are deterministic outputs of the governance chain replay.

This capability fundamentally changes the claims process. Disputes about whether controls were in place become verifiable rather than adversarial. Claim verification shifts from forensic reconstruction to cryptographic proof. The policyholder benefits because legitimate claims are validated faster. The insurer benefits because fraudulent claims are detected by the mathematics, not by investigators.

Premium Optimization

Organizations that maintain continuous attestation provide underwriters with a continuous risk signal. Control drift is detected the moment it occurs, not months later during an audit. This continuous visibility enables more accurate risk pricing. Organizations with strong attestation postures pay premiums that reflect their actual risk, not the average risk of their industry cohort.

Questions

Frequently Asked Questions

What is continuous attestation?

Continuous attestation is the practice of producing a cryptographic receipt for every control state change in an organization. Rather than sampling controls periodically, every modification to MFA policies, encryption configurations, access grants, key rotations, and agent permissions generates a post-quantum signed attestation that is independently verifiable and replayable.

How is continuous attestation different from continuous monitoring?

Continuous monitoring observes system state and generates alerts. Continuous attestation produces cryptographic proof of each state change. Monitoring tells you something changed. Attestation proves what changed, when it changed, under whose authority, and whether the change was policy-compliant at the moment it occurred. Attestation records are independently verifiable by any third party with a public key.

What control events does H33 attest?

H33 attests MFA state changes, encryption configuration modifications, key lifecycle events (generation, rotation, revocation), access policy changes, AI agent scope modifications, compliance posture transitions, data classification changes, and governance decisions. Each event produces a signed receipt containing the prior state hash, new state hash, authority chain, timestamp, and policy reference.

Why does attestation need to be post-quantum?

Attestation records must remain valid for years or decades. Classical signatures (RSA, ECDSA) are vulnerable to harvest-now-decrypt-later attacks: adversaries collect signed attestations today and forge them when quantum computers break the underlying mathematics. H33 signs attestations with three independent post-quantum signature families backed by three independent hardness assumptions. A forged attestation would require breaking lattice problems, NTRU structures, and stateless hash functions simultaneously.

Can insurers use continuous attestation as underwriting evidence?

Yes. Continuous attestation provides an unbroken evidence stream that underwriters can verify independently. Instead of relying on self-reported questionnaire responses, insurers receive cryptographically signed records of every control state change. The HATS protocol defines how attestation chains map to specific compliance controls, giving underwriters machine-verifiable proof that controls were active at any given moment.

How does continuous attestation work with governance replay?

Every attestation record is appended to a governance chain with deterministic ordering. Governance replay reads this chain and reconstructs the exact control state at any historical timestamp. Because each record includes the hash of the previous record and a PQ signature, the chain is tamper-evident and independently verifiable. Replay produces identical outputs regardless of who performs it or when.

Start Attesting Every Control Change

Deploy continuous attestation across your governance surface. Post-quantum signed. Independently verifiable. Every state change. Every receipt.

Get API Key Watch Demo