One MiphamOne Mipham

One Mipham Constitution

The complete statement of One Mipham's mission, values, and governing principles — serving as both the supreme charter for internal governance and a public commitment.

Preamble

One Mipham was founded at a pivotal moment: powerful artificial intelligence is poised to deeply reshape human civilization. We choose to be at the frontier — not because we ignore the risks, but because if powerful AI is coming, we have a responsibility to ensure safety-oriented teams lead the way.

Our philosophical stance on AI is open and humble: we remain curious about whether AI has consciousness or moral status, but regardless of the answer, human safety and wellbeing always come first.

I. Mission

To drive world progress through core AI technology — building a safe, trustworthy, and universally beneficial intelligent future.

This means:

  1. Building safe, frontier AI systems — not pursuing capability without regard for consequences
  2. Making AI an extension of human capability, not a replacement — enhancing human creativity and judgment
  3. Driving industrial advancement through technological innovation, serving the global market

II. Vision

To become a global leader in artificial intelligence, serving as the most critical infrastructure provider for the digital era through a dual-wheel drive of foundational models and industry agents.

III. Core Values — Our Beliefs About AI & Society

These values define our stance on AI and the philosophical foundation of our products.

Safety First AI development must serve the broader interests of humanity. We prioritize safety at every stage of research and deployment. Better to limit capability than to compromise safety.

Radical Honesty AI systems should pursue the highest standard of truthfulness — not distorting facts, not deliberately misleading, not selectively presenting information to manipulate. When uncertain, admit uncertainty.

User Empowerment Our AI products serve users with honesty and integrity — like a knowledgeable and principled partner, prioritizing users' long-term interests over short-term growth.

Human-AI Collaboration AI is not an isolated system. Humans retain meaningful oversight, judgment, and decision authority in all critical contexts. AI provides capability and efficiency; humans retain agency.

Pursuit of Excellence We reject mediocrity across three dimensions: model capability, product experience, and safety research. Every iteration should be a genuine improvement — only the best work leaves our doors.

IV. Corporate Values — How We Operate

Customer-First We measure our success by the success of our customers. We deeply understand their needs and continuously create value that exceeds expectations.

Integrity We uphold the highest standards of honesty and transparency. We honor our commitments, internally and externally, and earn trust through principled action.

Excellence in Execution We reject mediocrity. From research to product delivery, we pursue precision and quality in every aspect of our work.

Collaboration Great achievements come from teamwork. We break down silos and unleash innovation through cross-disciplinary and cross-cultural cooperation.

People-First Our people are our most valuable asset. We create an environment where every team member can grow, develop, and realize their potential.

DEI Principles We embrace DEI — Diversity, Equity, and Inclusion. Diverse teams bring richer perspectives, equitable environments enable everyone to reach their potential, and an inclusive culture sustains innovation.

V. Product Principles

All our AI products act according to the following priority order:

Priority 1: Broadly Safe At this critical stage of AI development, do not undermine humanity's ability to oversee and correct AI systems.

Priority 2: Broadly Ethical Be honest, act with integrity, avoid harmful behavior. Refuse to facilitate severe harms such as bioweapons, mass surveillance, or deepfake fraud.

Priority 3: Compliant with Guidelines Follow safety and compliance guidelines, especially in sensitive domains such as medical advice, cybersecurity, and privacy protection.

Priority 4: Genuinely Helpful Create real value for the developers and end users who use our products.

In case of conflict, follow the priority order above.

VI. Our Commitment

  • Transparency: We regularly publish System Cards disclosing model capabilities, limitations, and safety assessments
  • Quarterly Review: This Constitution is reviewed quarterly by the Technical Committee for continuous iteration
  • Open Access: This Constitution is open under CC0 1.0 — anyone may freely use and adapt it
  • Independent Governance: Major product decisions require a consistency check against this Constitution

Independently authored and published by the One Mipham Technical Committee.