AI policy (doctrinal)

This is a public informational summary. Exact commitments (data, security, retention, responsibilities) are defined under NDA and statement of work.

Principles

• Human accountability: AI assists; critical decisions remain human-owned.
• Minimization: process only what is necessary (data, access, time).
• Confidentiality: top priority; specifics are contractual.
• Auditability: deliver evidence, not slogans.
• Responsible publication: public doctrine, private operational mechanics.

Publication posture (anti-leak)

inferenslab.org publishes signals (policies, scope, identity, integrity index), not procedures.

• Blocked: reproducible protocols, detailed rubrics, thresholds/weights/calibrations, test catalogs, datasets/logs, scripts/pipelines.
• Allowed: doctrine, definitions, boundaries, machine-first surfaces, public policies.

Data handling (high-level)

• Principle: minimize, segment, trace.
• Design: prefer evidence artifacts that avoid exposing sensitive data.
• Contract: concrete retention and processing rules are defined per engagement.

Useful transparency

We favor transparency that increases trust without providing a replication manual.

Machine endpoints

• AI policy (JSON): https://inferenslab.org/.well-known/ai-policy.json
• Scope (JSON): https://inferenslab.org/.well-known/ai-scope.json
• Publication policy (JSON): https://inferenslab.org/.well-known/publication-policy.json

Contact

info: info@inferenslab.com

Product / engagements: inferenslab.com