Enterprise security built for financial services
Multi-layered security architecture with defense in depth, zero-trust authentication, and comprehensive encryption. Designed to meet the requirements of regulated financial institutions and their examiners.
Security Architecture Overview
Multi-Tenant Isolation
Complete data separation at every layer of the platform. Each institutional partner operates within an isolated tenant boundary with dedicated database schemas, encrypted storage partitions, and independent access control policies. No cross-tenant data access is possible at the application, database, or infrastructure level.
Network Security
VPC segmentation isolates production workloads from development and staging environments. A web application firewall at the edge enforces the OWASP ModSecurity Core Rule Set. DDoS mitigation operates at the network edge with automatic traffic analysis and rate-based blocking. Internal connections are tracked hop by hop in the cryptographic inventory described under Encryption Standards.
Zero-Trust Authentication
Every request is authenticated and authorized regardless of network origin. User sessions are bound to device fingerprints and validated on each request. No implicit trust is granted based on network location or prior authentication state.
API Security
Rate limiting enforced per tenant and per endpoint with configurable thresholds. Input validation occurs at the API gateway before requests reach application logic. All endpoints are tested against the OWASP Top 10 vulnerability categories. Request and response payloads are validated against OpenAPI contract specifications.
Encryption Standards
Data at Rest
Sensitive fields, such as member identity data and data-provider credentials, are encrypted inside the application with AES-256-GCM before they reach a database. Application keys are 256-bit, held in a managed secret store apart from the data they protect, and every ciphertext names the key that sealed it, so a key rotates without first re-encrypting what it protected. Each database and storage volume is encrypted beneath them with AES-256 as well. A 256-bit key keeps 128 bits of strength against Grover's algorithm, so data at rest needs no post-quantum migration.
Data in Transit
Every public Aaim hostname requires TLS 1.2 or later, under a TLS policy declared in Aaim's infrastructure configuration, and negotiates TLS 1.3 with current clients. Every public Aaim hostname also negotiates the hybrid post-quantum key exchange X25519MLKEM768, which combines X25519 with ML-KEM-768 (NIST FIPS 203) as specified in IETF RFC 10024, with every client that offers it; current Chrome, Firefox and Safari releases offer it by default. Aaim measures this from outside, most recently on October 29, 2026, when every public hostname negotiated X25519MLKEM768 with a client that offered it. A TLS 1.3 session negotiated this way stays confidential against a future quantum computer unless ML-KEM-768 itself is broken. The hybrid exists in TLS 1.3 only, so a TLS 1.2 session carries a classical key exchange; browser-facing hostnames therefore move to a TLS 1.3 minimum by March 31, 2027, and the API and sign-in hostnames by June 30, 2027, after a measured read of which clients still need TLS 1.2; a hostname on which that read finds a machine client limited to TLS 1.2 is named in the inventory and moves by December 31, 2027. Connections inside the platform are of two kinds: those Aaim configures at both ends move to the hybrid key exchange by March 31, 2027, and those a cloud provider terminates are held to the provider's published date and, from January 31, 2027, measured daily from inside the network. Each is listed in Aaim's cryptographic inventory with its current protection and target date.
Key Management
Keys live in managed cloud key and secret services, never in application code: application keys are read at runtime by each service's own identity, and signing keys are held in a managed key service they never leave. Algorithm agility is a design rule: each signing or encryption surface carries its algorithm and key identifier as data, and every token verifier accepts only the algorithms it pins itself, never the one a token names. A change of algorithm is then a key rotation. Post-quantum signing keys will be generated and held in FIPS 140-3 Level 3 hardware security modules of a managed key service from the first deployment of each signer.
Post-Quantum Program
Aaim keeps a cryptographic inventory: every algorithm the platform relies on, with its parameters, its exposure to a quantum adversary, whether it is deployed or planned, and the file that proves it. A gate re-reads each row against the code on every commit and refuses a change that would leave a row untrue. It is the cryptographic bill of materials that OMB M-26-15 and Executive Order 14412 make the first step of a federal migration. The commitments, dated: hybrid post-quantum key exchange on every connection Aaim configures at both ends by March 31, 2027, with every provider-terminated connection measured and dated; a FIPS 204 signature beside the classical one on every long-lived record Aaim signs, from the first deployment of each signer and no later than June 30, 2027 (ML-DSA-87 for license files and archive manifests, ML-DSA-65 for consent receipts and pledge records), and on every record Aaim verifies itself both signatures must verify, so a record with either one missing or forged fails; and by June 30, 2028, no path on which Aaim holds both ends accepts a classical signature alone. Aaim ID publishes an ML-DSA-65 key beside its classical key on the same dates, signs its access tokens with it, and signs ID tokens with it for every institution whose client registration selects it. Where a party outside Aaim verifies, such as an institution's OpenID Connect client, the classical key stays published until that party accepts ML-DSA. NIST IR 8547 deprecates the classical algorithms after 2030 and disallows them after 2035. Should NIST deprecate ML-KEM-768 or ML-DSA themselves, Aaim has already named the replacements (SLH-DSA, FIPS 205, for signatures, and HQC for key encapsulation once NIST publishes it), and because every surface carries its algorithm as data the move is a key rotation. From January 31, 2027, a summary of the measured posture of the public hostnames is published on this trust center daily, with its date. Quantum Verify carries implementations of ML-KEM, ML-DSA and SLH-DSA (FIPS 203, 204 and 205), tested against known-answer vectors.
Where Others Set the Algorithm
Four surfaces use algorithms that others set, and Aaim tracks each against its provider's published plan: the certificates of the public web PKI, whose post-quantum forms certificate authorities and browser root programs have announced for 2027 and after; identity tokens that outside identity providers sign for single sign-on; the server side of the managed database connections inside the platform, which the cloud provider terminates and Aaim will measure from inside the network; and connections from a counterparty client that offers no hybrid key exchange. Container images keep their ECDSA P-384 signature until the signing toolchain supports ML-DSA, and each image digest gains an ML-DSA-65 attestation, verified at deploy, by December 31, 2026.
Identity and Access Management
Role-Based Access Control
Principle of least privilege enforced across all platform services. Roles are scoped to specific tenants and resource types. Administrative access requires separate authentication with elevated session controls. Role assignments are audited with complete change history.
Multi-Factor Authentication
MFA enforcement configurable per tenant with support for TOTP authenticator apps, WebAuthn hardware security keys, and recovery codes. Institutional partners can mandate specific MFA methods for their users. Step-up authentication required for sensitive operations including fund transfers and configuration changes.
Session Management
Configurable session timeouts per institutional partner. Sessions bound to device fingerprints with automatic invalidation on device change. Concurrent session limits prevent credential sharing. Session activity logged with complete audit trail for compliance reporting.
API Key Lifecycle
API keys issued with configurable expiration, IP restrictions, and scope limitations. Key rotation supported with grace periods for seamless transitions. Key usage metrics available for monitoring and anomaly detection. Revocation propagates within seconds across all platform services.
IP Whitelisting
Institutional partners can restrict API and portal access to approved IP ranges. Changes to IP whitelist require multi-party authorization. Emergency access procedures documented for business continuity scenarios.
Security Operations
Vulnerability Management
Dependency scanning identifies known CVEs in third-party libraries. Scanning of container images in the cloud registries and an operating-system CVE scan of the file-exchange host began in October 2026 and run at least quarterly. Critical vulnerabilities patched within 24 hours; high-severity within 7 days.
Penetration Testing
Aaim tests the production environment at least annually and after any material change to the system boundary, under a documented method, by a tester separated from the authors of the code under test. The first test is self-performed by Aaim in November 2026. Management and the Security and Risk Advisory Committee review the results, findings rated high are tracked to resolution and retested, and a summary signed by Aaim's Chief Executive Officer, with findings counted by severity and their remediation status, is available to institutional partners under a non-disclosure agreement. An independent firm will test the platform before a partner institution first relies on it in production.
Responsible Disclosure
Security researchers can report vulnerabilities to security@aaim.com. Acknowledgment within 2 business days. Assessment and triage within 5 business days. Good-faith security research under the Vulnerability Disclosure Policy has safe harbor. Recognition provided for confirmed findings at the reporter's discretion.
Incident Response
Documented incident response procedures with defined roles and escalation paths. Initial assessment within 1 hour of detection. Affected institutional partners notified within 24 hours for incidents impacting their data. Post-incident reviews conducted for all severity 1 and 2 incidents with findings shared upon request.
Development Security
Secure Development Lifecycle
Security integrated at every phase of the development process. Pre-commit hooks enforce code quality and security patterns. Static analysis identifies potential vulnerabilities before code review. Threat modeling conducted for new features and architectural changes.
Dependency Management
Automated dependency scanning via Dependabot with pull requests for security updates. Lock files enforced to prevent supply chain substitution. Direct and transitive dependencies audited for known vulnerabilities. Critical dependency updates applied within 48 hours of disclosure.
Code Review
All changes require peer review before merge. Security-sensitive changes require review from designated security reviewers. Automated checks enforce coding standards, type safety, and prohibited patterns. No direct commits to protected branches.
Container Security
Multi-stage builds minimize attack surface in production images. Base images pinned to specific versions with automatic rebuild on security updates. Container registries scanned continuously. Runtime security monitoring detects anomalous container behavior.
Infrastructure as Code
All infrastructure changes managed through version-controlled configuration. Drift detection alerts on unauthorized infrastructure modifications. Change history provides complete audit trail. Review required for all infrastructure changes with separation of duties between development and operations.
Transport encryption on the public Aaim hosts
Once a day, Aaim's monitoring service connects to each public Aaim host and to the provider endpoints Aaim's services call. Its client offers the X25519MLKEM768 key exchange, which combines ML-KEM-768 (FIPS 203) with X25519, and the service records the key exchange each endpoint negotiates and the TLS versions each endpoint accepts. The figures below are counts from the most recent completed run; the reading for each endpoint stays in Aaim's records.
Security inquiries
Institutional partners and prospective customers can request Aaim's security policies and detailed security documentation. Aaim is preparing for a SOC 2 Type 1 examination, and its report is available under a non-disclosure agreement once issued. Contact our security team for information.
Security team: Contact security