Designed for the Frameworks
You Report Against
Solas Academy is designed to help regulated organisations meet their compliance obligations. Below is how the platform maps to major frameworks, what the differentiating controls look like in practice, and where formal certification work stands today.
Coverage at a Glance
Each row links to a one-paragraph explanation of what the coverage status means in practice for that framework.
EU AI Act
MappedArticles 9, 10, 13, 14, 15, 43, 47, 72Solas Academy is designed for high-risk AI system obligations under Regulation (EU) 2024/1689. Risk-management documentation under Article 9, data-governance and quality controls under Article 10, transparency to deployers under Article 13, and human-oversight workflows under Article 14 — including the Article 14(4)(b) human-override pathway for AI-graded learner submissions. Accuracy, robustness, and cyber-security controls under Article 15. Annex IV technical-file artefacts are maintained for conformity assessment under Article 43. Post-market monitoring per Article 72 lands as structured telemetry in the audit trail. The Article 47 Declaration of Conformity is held in draft until a first EU prospect requires signature.
NIST AI RMF
MappedGOVERN · MAP · MEASURE · MANAGESolas Academy is designed against the four functions of the NIST AI Risk Management Framework. GOVERN: documented AI policy, ownership, and accountability roles. MAP: AI use context, stakeholder identification, and impact assessment. MEASURE: model evaluation, telemetry on every AI grading call, bias and fairness signals. MANAGE: incident response, rollback paths, and the deterministic fallback that takes over when the AI grader fails closed.
ISO/IEC 42001
In ReviewAI Management SystemInternal alignment against the ISO/IEC 42001 AI Management System standard is in progress. The platform is built around the structural requirements — leadership, planning, operational control, performance evaluation, improvement — but the formal management-system implementation review is ongoing. Detailed gap documentation is available on request under NDA.
GDPR
Helps customers meetEU 2016/679Solas Academy is designed to help controllers and processors meet GDPR obligations: lawful basis tracking on enrolment, granular per-class retention policies, signed Data Subject Access Request (DSAR) exports, breach-notification telemetry, and field-level encryption at rest for sensitive identifiers. The hash-chained audit trail provides evidence of accountability under Article 5(2).
HIPAA
Helps customers meetUS HealthcareFor healthcare customers in scope of HIPAA, the platform is designed for administrative, physical, and technical safeguards: role-based access with read-only Compliance Officer separation, audit-trail completeness, MFA enforcement, encryption at rest, and configurable retention. Business Associate Agreement (BAA) execution is handled per deployment.
SOC 2 Type II
Helps customers meetTrust Services CriteriaSolas Academy is designed to support customers carrying SOC 2 Type II obligations across Security, Availability, Processing Integrity, and Confidentiality. Audit logs are cryptographically integrity-verifiable. Access controls, encryption, monitoring, and change-management evidence flow into the same export bundle used for other frameworks. Solas Cyber Security Solutions LLC does not itself hold a SOC 2 attestation today.
PCI DSS
Helps customers meetCardholder environmentsFor customers operating training programmes inside PCI DSS scope, the platform is designed for the security training and awareness requirements (Requirement 12.6 and related). Role-based assignment, completion evidence, and assessment results map directly to PCI awareness programme documentation.
ISO 27001
Helps customers meetISO/IEC 27001:2022Solas Academy is designed to map to the Annex A controls of ISO/IEC 27001:2022 most relevant to awareness, training, and personnel security — notably A.6.3 (information security awareness, education, and training). Module-to-control mapping is configurable so customers can produce evidence bundles grouped by their own control reference set.
NIS2
Helps customers meetEU Directive 2022/2555For essential and important entities under NIS2, the platform is designed to support the cyber-hygiene and awareness obligations (Article 21(2)(g)) and the personnel training elements of governance (Article 20). Evidence exports are structured for national competent-authority audits.
UAE PDPL
Helps customers meetFederal Decree-Law No. 45 of 2021For organisations in scope of the UAE Personal Data Protection Law, the platform is designed for data-subject-rights handling, consent management, breach-notification evidence, and retention controls. Regional deployment options keep personal data within UAE residency boundaries where required.
Three Controls That Make the Difference
These are the capabilities most awareness platforms do not ship by default — the ones an auditor will ask about.
Hash-chained audit logs with cryptographic integrity verification
Every administrative action, grading event, and policy change is appended to a hash-chained audit log. Each entry can be signed by the vendor Ed25519 key pair that signs licences and update bundles, so integrity is verifiable without trusting the database operator. Auditors can prove after the fact that no row was deleted or edited.
DSAR (data subject access request) support with audit-trail evidence
A subject’s record, completion history, certificates, quiz submissions, audit-log entries, and tutor conversations export as a single signed manifest. The manifest itself lands in the audit trail, so the customer can demonstrate both the request and the response under GDPR Article 15 and equivalent regimes.
Retention policies with automated enforcement and audit
Per-data-class retention is configurable at the organisation level — audit, completion, telemetry, and tutor conversations each carry their own period. Enforcement is automatic and recorded; an auditor can see when a class of data became eligible for deletion and when deletion executed.
How the AI Grader Stays Auditable
The short-answer grader is the most regulator-relevant surface in the platform, so it carries three layers of control that exist specifically to make Articles 13, 14, and 15 enforceable.
Post-parse rationale guardrail
Every grading response is parsed before any reasoning text reaches an admin or learner. When a prompt-injection or role-marker pattern is detected on the untrusted input boundary, the model’s rationale is replaced with a fixed system message. The model never speaks for itself on a flagged submission — the audit trail records both the trigger and the substitution.
Verdict-floor sanitiser
When adversarial input is detected, the grading verdict is force-failed regardless of what the model returns. Ambiguous parses fail closed; an eight-second timeout prevents an unresponsive provider from hanging a learner; on fallback the deterministic exact-match path takes over. Every call writes a telemetry row with prompt hash, tokens, cost, latency, finish reason, and guardrail triggers.
Article 14(4)(b) human-override workflow
Learners can request a re-grade. The request enters a queue where a human reviewer applies the rubric directly, overrides the AI verdict, and signs the decision into the audit trail. The original AI verdict, the rationale, the override, and the reviewer identity all remain visible to auditors.
Platform Security at a Glance
Hash-chained audit logs
Append-only, cryptographically integrity-verifiable. Ed25519 signing available against the same key that signs licences and updates.
Field-level encryption at rest
AES-256-GCM on SSO, SMTP, MFA, and webhook secrets. Optional per-tenant BYOK (AWS KMS or local-key id) where the customer prefers to hold the key.
MFA enforcement
TOTP, WebAuthn / passkeys, and backup codes on local accounts. SSO users authenticate via their identity provider (Azure AD / Entra ID, Okta).
Role-based access
SUPER_ADMIN → ORG_ADMIN → MANAGER → LEARNER, plus a read-only COMPLIANCE_OFFICER role designed for auditors and external reviewers.
Multi-tenant isolation
Single-tenant per deployment; multi-organisation capable inside one instance. Per-org IP allowlist (IPv4 + IPv6), per-org password policy with HIBP breach check, per-org concurrent-session cap.
Session and transport security
JWT 15-minute access + 7-day refresh with reuse detection. CSP hardened with no inline scripts; Helmet and CORS configured; Redis-backed rate limiting on auth endpoints.
Solas Cyber Security Solutions LLC does not itself hold ISO 27001, GDPR DPA registration, PCI DSS, HIPAA, or SOC 2 certifications today. The controls described above are platform capabilities Solas Academy ships, not certifications of Solas Security as a company. Roadmap items below describe the work in progress.
Detailed Documentation, Under NDA
Detailed compliance documentation — including Annex IV technical files, Article 9 risk-management documentation, Article 10 data-governance records, post-market monitoring artefacts under Article 72, and the draft Article 47 Declaration of Conformity — is available on request under NDA. Procurement, audit, and legal teams can request the full package directly.
Request Compliance DocumentationWhat’s Currently in Progress
Independent penetration testing
Engagement scheduled with an independent assessor. Findings and remediation will feed the security-evidence bundle described above.
ISO/IEC 42001 alignment
Ongoing internal review against the AI Management System standard. Gap analysis is available to evaluating customers under NDA.
EU authorised representative (Article 25)
Formal engagement of an EU authorised representative will be completed when the first EU prospect surfaces. Required before Article 47 signature.
Article 47 Declaration of Conformity
Held in draft, not signed. Will be executed alongside the EU authorised-representative engagement on the first EU deployment.
No dates committed publicly. Capability-tense framing intentional — status changes when the evidence is signed, not when work begins.
Questions on a Specific Control or Framework?
Compliance, procurement, and audit teams — reach the team directly. We answer control-by-control where it helps your assessment.
