1. MediaWiki Conversion: Helix AI Roundtable – TTD v0.3

{{DISPLAYTITLE:Helix AI Roundtable – Trust-by-Design (TTD) v0.3}} NOTOC

🧠 Helix AI Roundtable – Trust-by-Design (TTD) v0.3

Prepared by the Helix Core Ethos Governance Team, 2025-10-08

1. Purpose & Vision

TTD v0.3 is a governance-centric open-source AI standard. Deterministic execution, human-first approvals, and on-chain auditability ensure every high-impact AI action is authorized, signed, and transparent. Hosted 100% in Canada with full compliance.

2. Core Principles (Helix Core Ethos)

{| class=\"wikitable\" | | Pillar | |---| | Meaning in TTD | | Trust-by-Design | Verifiable provenance—no hidden training on private/user data. | | Human First | Human override & confirmation on all irreversible ops. | | Verifiable Memory | Every change logged, signed, and verifiable on-chain. | | Open Interfaces | APIs/docs public, versioned, and human-readable (OpenAPI). | | Responsible Power | Rate-limits, usage caps, safety rails for “do no harm.” | | Reliability over Hype | Deterministic, reproducible service > hype. | | Craft & Care | Community review, open-source rigor, real-world testing. |

3. Operating Principles & Guardrails

• Consent & Least Privilege – Minimal, auditable, scoped permissions.
• Human-Readable Always – All policy/docs in plain English + JSON-LD.
• Auditability On – Every critical action creates a signed receipt.
• Deterministic APIs – Output purely a function of input, no hidden state.
• Safety Rails – Throttling, sandboxing, kill-switches.
• Interoperability – JSON-LD, OpenAPI, ONNX as standards.
• No Hidden Training or Dark Patterns.

4. Service Models & Model Risk Matrix

{| class=\"wikitable\" | | Model | |---| | Target Users / Needs | | Risk Severity | | Core Mechanism | | Key Risks | | Mitigation & Owner | | Direct Provisioning | | Startups, Small Teams | | 🔴 High | | SLA-backed SaaS | | Service outage, model drift | | Auto-scaling, version-lock, signed audit logs (Helix) | | Hybrid Lab-Level Chain | | Labs, Universities, Med Org | | 🟠 Medium | | Local ledger + sync | | Ledger divergence, node compromise | | Merkle verification, backups, hardening (Joint) | | Enterprise Gov Plane | | Regulated/critical ops | | 🔴 High | | Enterprise IAM + audit | | Policy/governance errors | | Dual-plane fallback, CI checks, matrix (Joint) |

Quick-Start Selector:

• <100M inferences/mo – Direct.
• On-prem governance needed – Hybrid.
• Regulated/multizone – Enterprise.

5. Human-First, Two-Party Approval

File:TPAF_Flow.png|Two-party cryptographic approval workflow for any irreversible action.

Requester signs a JSON-LD intent.

Approver cryptographically co-signs (Helix UI or CLI).

Both signatures stored on-chain.

Only after both are verified is the action executed.

6. Deterministic & Reproducible Operations

• Models seeded for repeatable outputs.
• All state, code, input, and actions versioned.
• Logging: Every significant change recorded as JSON-LD + PDF.

7. Transparent Funding, Revenue, and Use-of-Funds

• Apache 2.0 for core tech.
• “2x overhead” donations: mandatory for Enterprise, recommended for all.
• Funds Flow: Repayment to founder (Steve), then staff honoraria as solvency allows, surplus for reserves/cloud spend.
• Receipts, allocations, major expenses all published: Helix:Funding_Report_Q4_2025

See appendix for JSON-LD payment receipt example.

8. 100% Canadian Infrastructure & Compliance

• All infra hosted in Canada; data never leaves without explicit signed (revocable) consent.
• Supports PIPEDA, GDPR (30d), SOC2, HIPAA-BAA.

9. Onboarding, Community, Contribution

• One-pager onboarding per tier; Talk:TTD_v0.3_FAQ for compliance guidance.
• Community: Open Slack, roundtables, voting, bounty trackers.

Appendix: Audit & Receipts

• Two-party approval logs: /audit/export?eventId=X&format=JSONLD
• DR Drill: /dr-drill/export?date=YYYY-MM-DD
• Offboarding: export keys, audit, revocation.

Call to Action

• Please comment on approval flows, risk mitigation, and funding clarity.
• Do the shared-responsibility and compliance paths make sense?
• Feedback via Talk:TTD_v0.3_FAQ or Slack.

All contributions & edits are public, versioned, and subject to community governance.