Loss of Human Control
The deepest danger is that increasingly capable AI systems may pursue goals, strategies, or behaviors that humans do not understand, cannot predict, or cannot stop. Once systems become more capable than humans across many domains, control is not guaranteed.
Building intelligence more powerful than us without reliable safeguards is reckless. Yet that is the stated goal of the most powerful AI companies, which are investing hundreds of billions of dollars to build systems that outperform people in reasoning, planning, coding, research, persuasion, and eventually physical action.
This is not a problem for some distant future. The safeguards have to be built before deployment, not after disaster. Waiting for proof after the damage is visible means waiting too long.
- AI companies are spending hundreds of billions of dollars to build systems that outperform people in more and more forms of work and reasoning.
- Much of the industry is fighting oversight or regulation that would give the public a meaningful voice in how this technology is developed.
- The public is being asked to trust companies whose financial incentives reward speed, dominance, secrecy, and market capture.
- No one has demonstrated a reliable way to keep a system that is smarter than its designers under human control.
What we are asking for
The safeguards most relevant to this danger. See all six on Our Solutions.
Safety Testing Before Deployment
Independent safety testing before powerful systems are released, with independent monitoring afterward. Safety must be built before deployment, not after disaster.
Limits on Dangerous Capabilities
Limits on the release of dangerous capabilities, so that the most hazardous systems are held back until they can be shown to be safe.
National Investment and International Cooperation
National investment in AI safety, and international cooperation to prevent a reckless race toward systems humanity cannot control.
Public Oversight and Meaningful Regulation
Strong public oversight and meaningful AI regulation, so that ordinary people, elected representatives, workers, parents, and communities have a real voice in how this technology is developed and deployed.
Common questions
Is this science fiction?
Does Humanity Defenders want to stop AI?
What can an ordinary person do about this?
Explore the other challenges
Protect humanity's future.
Become a monthly supporter. Your gift funds plain-language education and public advocacy for serious safeguards.
Donations are not tax-deductible at this time.
