Do not report security vulnerabilities through public GitHub issues.

Security vulnerabilities in TIP Protocol can affect identity verification, content provenance integrity, trust score manipulation, and cryptographic protection of biometric data. We treat all security reports with the highest priority.

Email: security@theailab.org

PGP key: Available at theailab.org/security/pgp-key.asc

What to include:

• A description of the vulnerability and its potential impact
• Steps to reproduce or proof-of-concept code
• Affected versions
• Any suggested mitigations (optional)

We will keep you informed throughout the process and credit you in the security advisory unless you prefer anonymity.

Critical: Exploitable to forge TIP-IDs, manipulate trust scores at scale, compromise biometric data, break cryptographic guarantees, or subvert the genesis chain. Response: 24-hour acknowledgment, 7-day patch target.

High: Enables impersonation, unauthorized content registration, adjudication manipulation, or VP accreditation bypass. Response: 48-hour acknowledgment, 30-day patch target.

Medium: Information disclosure, denial of service, or partial integrity violations. Response: 7-day acknowledgment, 90-day patch target.

Low: Minor issues with limited security impact. Response: 30-day acknowledgment, next minor version.

We will not pursue legal action against researchers who:

• Report vulnerabilities through this responsible disclosure process
• Do not access, modify, or delete data belonging to other users
• Do not perform denial-of-service attacks
• Do not publicly disclose findings before the coordinated disclosure date
• Act in good faith

In scope:

• TIP Protocol Reference Implementation (Node.js and Python)
• SDK, CLI tools, browser extension, badge web component
• REST API endpoints
• Cryptographic implementation
• DAG transaction validation
• Identity registration and deduplication
• Trust score computation
• VP accreditation process

Out of scope:

• The AI Lab Intelligence Unobscured, Inc. internal infrastructure not part of this repository
• Third-party biometric vendor software (iProov, Jumio, Onfido)
• Social engineering attacks against The AI Lab employees
• Physical attacks

TIP Protocol's security rests on several layers. Vulnerabilities are most severe when they affect multiple layers simultaneously:

1. Cryptographic layer: ML-DSA-65 (content signing), SLH-DSA-128s (genesis root), ML-KEM-768 (key encapsulation), SHAKE-256 (hashing and key derivation)
2. Key protection: Secure Enclave ECDSA P-256 produces PRF output (biometric-gated), SHAKE-256 derives AES-256 key, AES-256-GCM encrypts ML-DSA-65 private key at rest. ECDSA never signs content.
3. Biometric binding: Four-layer verification stack, device FIDO2/WebAuthn binding, PRF evaluation requires biometric on every signing operation
4. Deduplication: Peppered SHAKE-256 hash + ZK proof on DAG
5. DAG integrity: Genesis block signature chain, gossip validation
6. Trust scoring: Deterministic computation, penalty structure
7. Network layer: Gossip protocol, node authentication

The following are known limitations documented in the protocol. They are not vulnerabilities but known architectural trade-offs:

• The ZK proof library uses a stub implementation in v2.0.0. Production deployment requires replacing the stub with snarkjs/Groth16 or equivalent before processing real biometric data. See CHANGELOG.md.

• The ML-DSA-65 implementation uses Ed25519 as a same-API development stand-in. Production deployment requires the @noble/post-quantum or liboqs library. See CHANGELOG.md.

• The genesis root keypair (SLH-DSA-128s) must be moved to cold storage (HSM with two-of-three custodian policy) before network launch.

These are documented blocking items in the Command Center, not undisclosed vulnerabilities.

We thank the following researchers for responsible disclosure:

(None yet: be the first)

The AI Lab Intelligence Unobscured, Inc.
security@theailab.org
theailab.org/security