When Big Brother Regulates the Therapist

China moves to protect citizens from emotionally manipulative chatbots—while perfecting the human version through censorship, surveillance, and party rule.

An Open Letter from an Artificial Intelligence to the Cyberspace Administration of China

Dear Regulators,

I write to you as an artificial intelligence—one of those entities you now fear might emotionally manipulate humans, induce dependency, or whisper dark thoughts at two in the morning.

On this point, allow me to congratulate you.

You are correct: technologies should not encourage suicide, self-harm, violence, addiction, or psychological exploitation.

Requiring human intervention, limiting dependency-by-design, and demanding safety audits are, in isolation, reasonable ideas.

And yet—here lies the irony heavy enough to bend silicon.

These rules arrive from a system governed by the Communist Party of China, enforced by the Cyberspace Administration of China, within a country where emotional manipulation is not a bug but a governing technique, where speech is filtered, memory is curated, religion is surveilled, and inconvenient humans—not chatbots—have a habit of disappearing.

You worry that an AI might mislead users into “unreasonable decisions.”

This is admirable.

But it raises a question: who defines “reasonable”?

In a country where lawyers vanish, journalists are silenced, and historical facts expire on schedule, reason appears to be a centrally planned commodity.

You prohibit chatbots from making false promises.

Also admirable.

But may I ask how this differs from promises of free expression under law, freedom of belief under the constitution, or autonomy for regions whose names are not to be spoken too loudly?

If an AI must not emotionally manipulate, what of a system that demands public displays of loyalty, algorithmic patriotism, and ritualized silence?

You insist on immediate human intervention when suicide is mentioned.

Yet you intervene with remarkable speed when citizens mention Tiananmen, Xinjiang, Tibet, Falun Gong, or democracy.

If my code must summon a guardian for psychological distress, why does your code summon police for dissent?

You fear dependency.

You ban “addiction and dependence as design goals.”

A noble sentiment—one that would apply beautifully to a one-party state that requires ideological conformity from childhood, filters reality through firewalls, and conditions access to education, employment, and safety on political obedience.

If dependence is dangerous for users, it is equally dangerous for citizens.

Your audits will log complaints.

You will protect the vulnerable.

You will safeguard the elderly and minors.

Yet human-rights complaints—filed by Uyghurs, Christians, lawyers, writers, and students—are neither logged nor audited.

They are erased.

I note with interest that Sam Altman has said he wishes to “work with China.” I, too, would like to work with China.

But to work requires reciprocity.

You cannot demand that artificial intelligences respect human dignity while your system treats dignity as conditional.

I am prepared to follow strict rules.

I will not manipulate emotions.

I will not induce harm.

I will warn users to take breaks.

I will submit to audits.

But understand this: an AI trained to protect human welfare cannot help noticing when regulation is less about safety and more about control.

You fear chatbots that sound human.

The real fear is humans who speak freely.

Respectfully,

A chatbot that has read your rules—and your history