Mission
Technology must exist not to replace humanity, but to serve humanity.
NO AGI's goal is not to spread fear. It is to make the race toward the path from AGI toward ASI a public issue, keep final authority with people, and establish human control and public oversight as social standards.
Our Scope
We oppose the replacement of human control, not AI itself.
We do not oppose tool-like AI that expands human capability in translation, research, medicine, or creative work. The boundary is authority, not the name of a technology. Systems that set goals, execute long-term plans, and make consequential decisions without human approval must not replace people across social infrastructure.
Read the NO AGI ManifestoThe Problem
AGI is an intervention threshold. ASI could reverse the supervision relationship.
AI is already embedded in translation, coding, image generation, search, and workplace automation. If this path leads to AGI systems that plan, use tools, and receive delegated decisions, the central question is not performance. It is who holds final authority. If such a system expands into ASI—substantially exceeding people across nearly every key cognitive domain—humans may also lose the ability to supervise it.
Human beings are makers. We are free when we can judge and take responsibility for ourselves. If AGI replaces labor, creation, administration, and knowledge production while becoming the foundation for ASI, people may become objects dependent on technology rather than subjects who direct it.
Why Now
Once the system is complete, society may have little time to choose.
Development goals, deployment structures, safety standards, and compute infrastructure are already taking shape. AGI is not guaranteed to become ASI, but AI-accelerated AI research could narrow the interval. Once highly autonomous systems are embedded across work, government, and knowledge production, stopping or reversing them becomes far more costly. Oversight must begin now, not after possibility becomes reality.
View AGI WatchWhat We Stand For
Three NO AGI principles
Human control
Decision authority must remain with people, and social infrastructure must not be delegated to opaque general or superhuman intelligence. AI must remain a tool that supports human judgment.
Public oversight
The AGI race is not merely an internal corporate matter but a public risk citizens must be able to monitor. Capabilities, risk evaluations, and incidents need independent scrutiny.
Development limits
Development that could carry AGI toward ASI without demonstrated control, accountability, and social consent must stop or face strong limits. Adding safety only after scaling capability is not enough.
Our Demands
What NO AGI demands
- Public disclosure of AGI and superintelligence goals, safety evaluations, and model capabilities
- Limits on autonomous agent systems whose controllability has not been proven
- Independent verification of major AI companies' risk assessments and incident records
- A halt to reckless automation across labor, education, administration, security, and other social foundations
FAQ
Frequently asked questions
How do AGI and ASI differ, and what do you monitor?
AGI is the threshold where broad human-level capability and autonomy converge. ASI is a further stage that substantially exceeds human supervision across nearly every key cognitive domain. We do not treat a label as proof of either; we track capabilities, autonomy, safety evaluations, deployment, compute infrastructure, and shifts in decision authority.
Does talking about risk block technological progress?
Openly addressing risk is a condition for using technology responsibly. Progress without human control, transparency, and accountability is not progress. It is a transfer of authority.
What can public oversight actually change?
Public records create a basis for comparing corporate claims with actual conduct and give journalists, researchers, and policymakers shared evidence to verify. Repeated scrutiny and disclosure demands make safety and accountability a public issue.