People get access.
People can use the system. The provider still controls the runtime, policy, data, and default placement.
The AI future is for everyone
Broad access can put powerful AI in every hand. Access alone does not decide who sets the rules, who runs the system, or who gets to say no.
The first graphic shows access. Keep scrolling to see what access does not guarantee.
Each dot represents access, not a person or a measured adoption rate.
Access, control, consent
Start with access. Advance the scrubber to add control, then consent.
People can use the system. The provider still controls the runtime, policy, data, and default placement.
Running an open-weight model on your own infrastructure gives you control of the runtime. It does not give you control of training or guarantee a non-AI path.
Consent needs a clear opt-in or opt-out, a choice that lasts, and a product that still works without the model.
The essay offers three principles. The first two describe what people might build. The third asks who should control the capability.
Give people powerful tools to pursue the life and work they choose.
Give people access to tools for businesses, discoveries, treatments, and new work.
Give more people access instead of concentrating control inside a few institutions.
The essay points to open-source history in cybersecurity, then makes room for stronger coordination around biological risk. Someone still has to set the threshold.
Broad access can spread knowledge, expose weaknesses, and help defenders repair them. This is the essay’s strongest historical analogy.
Ask whether a person has access, control, and a real choice. Each answer depends on how much of the system they can run.
“Access is not agency. Distribution is not consent.”
The next story asks whether everyone includes the person who does not want it.