Skip to content
OpenAGI

#11 · Manifesto

OpenAGI

Open Standards for Safe AGI Development

"The race to AGI shouldn't be a race to the bottom on safety. We're building the open standards, compliance frameworks, and certification programs that make responsible AGI development the competitive advantage."

THE REFUSAL

We refuse to accept that AGI safety standards must come from governments who don't understand the technology. We refuse to accept that voluntary commitments without verification are meaningful. We refuse to accept that competitive pressure must erode safety practices. We refuse to accept that responsible labs have no way to differentiate themselves from reckless ones.

THE INSIGHT

The AGI safety landscape is fractured. Every lab has internal guidelines. Governments propose regulations they can't enforce. Academics publish principles no one implements. What's missing is the connective tissue: shared standards with shared assessments and shared accountability.

Industries facing existential technology risks have solved this before. Nuclear power has the NRC and IAEA. Aviation has the FAA and ICAO. Finance has Basel accords. AGI needs its standards body—and we're building it before governments build something worse.

THE AUDACIOUS BET

Within 5 years, OpenAGI certification will be the recognized standard for responsible AI development. Labs without certification will face market and regulatory pressure. Labs with certification will have trusted third-party validation of their safety claims.

WHY NOW

SURVIVING THE THREE DEATHS

Law 1: Expect the Three Deaths

RELEVANCE DEATH

The threat: If the major labs don't participate, the standards are meaningless.

How we survive: We're building with lab input from day one. Our standards reflect technical reality because practitioners write them. We're not imposing external constraints; we're codifying internal best practices.

CAPTURE DEATH

The threat: Industry standards bodies often become tools of incumbents to block competition.

How we survive: Our governance includes diverse stakeholders: labs, academics, civil society, affected communities. No single voice dominates. Transparency is mandatory.

ENFORCEMENT DEATH

The threat: Without enforcement power, certification is just theater.

How we survive: We don't need legal enforcement. Market pressure (investor requirements, enterprise procurement, regulatory reference) creates economic enforcement. We make responsibility the rational choice.

ARTISTS & SOLDIERS

Laws 2 & 3: Love Both Equally, Manage the Transfer

Our Artists: AI safety researchers defining what 'aligned' means technically, governance scholars designing institutional structures.

Our Soldiers: Compliance specialists who understand certification programs, auditors with AI expertise, communications teams building public trust.

The Transfer Mechanism: Quarterly 'Standards Summits' where researchers propose technical criteria and compliance experts evaluate audit feasibility.

STRUCTURAL MOAT

Law 4: Create Innovative Structure

SUCCESS METRICS

Law 5: Gardeners Measure Growth

North Star: Labs and models certified under OpenAGI standards

Leading Indicators: Standards contributors, Assessment completions, Evidence artifacts submitted

Business Metrics: Certification revenue, Training/consulting revenue, Grant funding

THE TEAM WE NEED

We need AI safety researchers who can define technical standards. We need governance experts who understand standards body design. We need diplomats who can build consensus among competing interests. We need communicators who can build public trust.

THE CALL

AGI will be the most transformative and potentially dangerous technology ever created. We're making sure its development has guardrails.

OpenAGI — An Acelera Intelligence Loonshot