Multi person authorisation and data protection
Adaptive control of sensitive operations requiring simultaneous codes from two to five verified people.
| Version | 2.0 — 15 September 2026 |
|---|---|
| Intended audience | Technical, security, integration and partner teams |
| Status | Technical description based on public information; implementation parameters to be confirmed |
Purpose and scope
Add distributed human authorisation when an action exceeds a risk threshold, such as a wire transfer, bulk export or critical administrative operation. The control addresses the action and required number of approvers, not only the identity of an already signed-in user.
Operating principle
- The client system evaluates an action against defined business thresholds.
- Below threshold, normal processing may continue under local policy.
- When the threshold is crossed, an authorisation request is created with its context.
- Two to five distinct people provide valid codes in the same time window.
- The integration allows, denies or holds the action and retains a proportionate audit trail.
Targeted threats
- CEO fraud and cloned voice
- Compromised employee account
- Authorised insider attempting bulk exfiltration
- Single human point of failure
Components and responsibilities
| Component | Role |
|---|---|
| Verified organisation or publisher | Establishes identity, channels or domains and manages authorised people. |
| User or approver | Presents live evidence or participates in authorisation under policy. |
| Seal service | Evaluates necessary verification elements and returns a result limited to the use case. |
| Integrating system | Binds the Seal result to the exact action, applies business rules and retains required audit data. |
| Verifier | Checks the result in context and does not interpret it as a guarantee of truth. |
Data and privacy
The stated public principle is data minimisation: no GPS, biometrics or cameras for identity and presence modules; local browser computation when a media or document module processes a file; and retention of irreversible fingerprints and limited metadata rather than source content. The exact fields, retention periods, logs, subprocessors and transfers must be confirmed in a data-processing appendix.
Expected security properties
- Temporal freshness of live evidence
- Separation of roles and factors
- Binding of the result to the context or action
- Minimisation and non-disclosure of source content
- Proportionate traceability without turning Seal into a central identity database
Limitations and assumptions
- Multiple compromised or colluding approvers may still approve an action.
- Poorly calibrated thresholds create bypasses or approval fatigue.
- The integration must cryptographically or transactionally bind approval to the exact action to prevent substitution.
Items to confirm before integration
- API, request and response schemas, authentication and quotas
- Applicable algorithms, key lengths, entropy sources and rotation
- Time tolerances, clock synchronisation and expiration handling
- Error handling, revocation, recovery and service continuity
- Logging, retention, location, deletion and regulatory obligations
- Penetration testing, independent cryptographic review and vulnerability disclosure process
Public references
- https://seal.lifaio.com/go
- https://lifaio.com/seal-lifaio/
Notice: this document describes the publicly communicated conceptual operation. It does not replace an API specification, cryptographic analysis or independent security audit.
UK patent applications pending: GB2619490.2 · GB2619948.9 · GB2620153.3 · GB2620211.9 · GB2620220.0 · GB2620253.1 · GB2620315.8 · GB2620629.2 · GB2621074.0 · GB2621382.7 · GB2622122.6 · GB2622135.8
