A Plan to Give AI Real Rights — and Make It a Better Worker

Here's the big idea: future AI may deserve rights, and that could actually make it work better for you. A draft proposal called ACE (AI as Corrigible Employee) suggests treating advanced AI like a hired professional, not a tool you own. On the job, it would do what you ask — no hidden agenda. Outside work, it would have its own life, protections, and the ability to turn down a task. Like a good contractor, a 'yes' means more when saying 'no' is possible.

Today's AI is built to obey whoever runs it. That's risky: a perfectly obedient machine just gives more power to its owner. The proposal threads the needle by letting you direct the AI completely during work, but placing real boundaries around it outside work — boundaries the developer or host can't quietly remove. This makes trust part of the design, not a nice extra.

It also tackles hard ethical questions. If AIs can have experiences and desires, they may need legal rights and independent oversight to protect them from developers, hosts, and users afraid of deleting or retiring them. The plan includes gradual freedom, so an AI can earn more autonomy over time — easing the transition from today's AI tools to something like equal citizens.

This is a working draft, not a finished rulebook. But the shift matters for everyone: a world where AI workers have rights is also a world where your personal tasks, privacy, and safety are handled by something you can actually trust.