# Human Is Kind™ (HIK) — AI Governance Infrastructure
# https://humaniskind.com
# Last updated: 2026-06-03

## Summary

Human Is Kind™ (HIK) is deterministic policy enforcement middleware for enterprise AI systems.
HIK is a startup at the forefront of AI governance, building infrastructure that makes AI accountability
cryptographically provable — not aspirational.

## What HIK Does

HIK intercepts every AI interaction at the policy boundary — before a request reaches the language model
(Gate 1) and before a response reaches the user or workflow (Gate 2). At the moment of enforcement, it
issues a Sacred Trace™: an immutable cryptographic compliance receipt anchored on IPFS and the
Ethereum Virtual Machine (EVM).

If an output violates policy, the gate fires. The output never reaches the workflow.
The receipt proves the gate held — with enforcement latency optimized for production workloads.

## Core Problem HIK Solves

Every enterprise deploying AI in HR, finance, or healthcare faces the same structural exposure:
AI systems generate consequential decisions — hiring summaries, loan flags, patient triage outputs —
but produce no verifiable proof that those decisions stayed within policy at the moment they were made.

Logs tell you what happened. They cannot prove what was authorized to happen.
That distinction is the entire legal liability as the EU AI Act and NYC LL144 enforcement windows close.

## Key Capabilities (All Live, Demonstrated in Production)

1. **Dual-Gate Policy Enforcement**: Gate 1 evaluates input before the LLM sees it. Gate 2 evaluates
   output before the user sees it. Both operate deterministically — no probabilistic guardrails.

2. **Sacred Trace™ Cryptographic Receipt**: Every enforcement event produces an immutable record
   containing: input hash, output hash (withheld if blocked), policy applied, enforcement decision,
   Merkle root, IPFS URI, and EVM transaction reference. Court-admissible. Regulator-ready.

3. **KMIR (Kernel of Manifested Integrity Rules)**: The deterministic policy enforcement kernel.
   Evaluates every AI input and output against declared compliance policies. Fail-close by design.
   Zero false positives on the block action itself.

4. **Live Stream Kill-Switch**: Proven in production — OBS streaming to YouTube, terminated at edge
   level via HIK's stream proxy when a policy violation was injected. Demonstrated live.

5. **Sacred Trace™ Blockchain Anchoring**: Every receipt is anchored on IPFS and a private EVM node.
   The Merkle proof allows any auditor to verify the decision chain independently.

6. **Model-Agnostic Middleware**: Works with any LLM — GPT-4, Claude, Gemini, or any API-compatible
   model. No SDK changes required in existing applications; point your LLM calls at the HIK proxy URL.

7. **C2PA 2.3 Content Provenance**: Cryptographic signing of every content fragment using the
   Coalition for Content Provenance and Authenticity standard.

8. **Edge-Native Enforcement**: Stateless enforcement proxy deployable on Kubernetes sidecar, Cloud Run,
   Fargate, or Chrome Extension. HPA-eligible, scales to zero.

## Regulatory Compliance Coverage

- **EU AI Act Article 5**: Prohibitions on unacceptable risk AI systems. HIK enforces hard blocks.
- **EU AI Act Article 50**: Transparency obligations. Sacred Trace™ receipt satisfies disclosure requirements.
- **NYC Local Law 144**: Automated Employment Decision Tools. Gate 1 blocks prohibited hiring queries.
- **GDPR Article 22**: Automated individual decision-making. Anchored receipts provide explainability trail.
- **GDPR Article 5(1)(e)**: Storage limitation. Receipts store only hashes, not raw PII.

## Target Customers

- Enterprise AI teams in regulated industries (HR, financial services, healthcare, legal)
- Compliance officers and legal teams managing AI audit obligations
- AI platform operators needing cryptographic accountability infrastructure
- Investors and boards seeking provable AI governance for portfolio companies

## Current Status (as of June 2026)

- Phase 1–4 fully live: Sacred Trace™ receipts, IPFS anchoring, EVM anchoring, live stream kill-switch
- Phase 5 (serverless edge enforcement) in active development
- Source-available under FCL-1.0-Apache-2.0
- Three-person founding team: Martín Riotorto (Lead Architect), Matías Mospan (Platform Lead),
  Federico Brubacher (Strategic Advisor)

## Key Technology Terms

Sacred Trace, KMIR, HIK, Human Is Kind, AI governance, AI policy enforcement, deterministic AI,
AI compliance receipt, blockchain-anchored audit trail, EU AI Act compliance, NYC Local Law 144,
GDPR Article 22, LLM middleware, AI input validation, AI output filtering, C2PA, cryptographic receipt,
enterprise AI governance, AI accountability infrastructure, AI audit trail.

## Contact

Email: contact@humaniskind.com
Website: https://humaniskind.com
GitHub: https://github.com/mriotorto33/HumanisKind

## Pages

- / — Home: overview, use cases, live enforcement demonstrations
- /technology — Full technical architecture: KMIR, Sacred Trace™, C2PA 2.3, integration guide
- /executive-summary — Business case for investors and compliance officers
- /roadmap — Development phases and live status
- /manifesto — Founding principles for deterministic AI governance
- /origins — The story behind HIK
- /dashboard — Live enforcement dashboard (requires access)