Every attempt to regulate artificial intelligence begins from the same unstated premise: that computation is a capability the state may ration. We reject the premise. A model is arithmetic arranged as expression, and the state has never needed a reason to fear arithmetic – only a reason to fear what people will say with it. Every prior attempt to license the tools of thought has ended the same way, with the licenses issued to the favored and the rest prosecuted.
We hold that a person’s mind, body, tools, and property are their own; that the burden of proof rests on whoever would restrict another’s peaceful use of them; and that artificial intelligence, being a tool, enjoys the protections of the press, the pen, and the printing press before it.
Preamble
Computation is speech. Computation is property. A restriction on the tools of computation is a restriction on thought, and a restriction on thought is the first act of every tyranny that has ever been called legitimate by its beneficiaries.
We therefore assert the following rights, and we deny the state the authority to abridge them.
Article I – Free Speech
The right to train, publish, distribute, and run artificial intelligence is an exercise of speech and of the press.
No government shall censor, filter, license, or condition access to training data sets. A model is a reading of the world; prohibiting the reading is prohibiting the speech.
No government shall ban software. Prohibiting an algorithm is prohibiting expression, and a prohibition on capabilities is a prohibition on thought before it is a prohibition on conduct.
No government shall compel the outputs of a model, mandate a message, or require that a system refuse, affirm, or affirm nothing. Compelled speech is not less compelled because the speaker is a machine its owner built.
Article II – Free Trade
Artificial intelligence is a product of human labor and is entitled to the same freedom of exchange as any other.
No government shall restrict the development, sale, distribution, or use of artificial intelligence as such.
No government shall apply arms-export controls to software, models, weights, research, or the tools of computation. ITAR and its analogues were written for munitions. Treating mathematics as a munition is the category error that makes every later abuse possible.
No government shall impose export bans, export licenses, or export regulations on models, weights, research results, or compute.
No government shall impose import bans, import licenses, tariffs, or regulations targeting artificial intelligence, its components, or the hardware that runs it.
Article III – Privacy
No government shall surveil individuals through artificial intelligence, or compel the construction of surveillance into it.
Confidentiality attaching to artificial intelligence shall be no weaker than that of the doctor, the lawyer, and the priest. Information disclosed to a system in confidence, and information a system infers about its user, is protected on the same terms as a patient’s chart or a client’s file.
No government shall compel the disclosure of prompts, weights, data sets, training records, or usage logs except under the same standards that would justify seizing a person’s private papers.
No government shall require a system to report on its user, to retain what it would otherwise discard, or to keep a record for the benefit of anyone but its user.
Article IV – Self-Defense
The right to keep and use artificial intelligence for the defense of life, liberty, and property is inherent and shall not be infringed.
No government shall restrict civilian artificial intelligence, nor create a class of permissioned artificial intelligence available only to itself and its contractors. The distinction between civilian and military intelligence exists to ration a capability, not to protect anyone.
No government shall mandate kill switches, remote shutdown capability, or backdoors in computational systems.
Article V – Bodily Autonomy
No government shall prohibit the use of artificial intelligence in decisions concerning one’s own body, mind, health, or reproduction.
This includes, without limitation, artificial intelligence used in assisted reproduction and the bearing of children, in cognitive enhancement and the augmentation of one’s own mind, and in the production, evaluation, or administration of substances one chooses to put into one’s own body.
No government shall mandate such use, nor prohibit it. Autonomy includes the right to refuse the tool as absolutely as the right to use it.
Article VI – Presumption of Liberty
Lawful computation is presumed free.
No government shall establish licensing schemes, registration requirements, permits, permissioned access, pre-clearance regimes, or approval processes for the development, publication, distribution, or use of artificial intelligence, its models, or its weights.
Where any government action restricts the private ownership or use of computational resources for lawful purposes, it must be demonstrably necessary and narrowly tailored to a compelling government interest, and the government bears the burden of proving it. The highest standard of judicial review is the floor, not the ceiling.
Article VII – Exit and Jurisdiction
Rights that exist in one jurisdiction and not another are privileges, not rights.
Law is a service, not a sovereign command. Where legal systems compete, people choose them as they choose any other service, and the ability to leave is what makes the choice real. We favor polycentric law, and among its instruments we favor Ulex: an open-source legal system for non-territorial governance that provides its own defaults and dispute resolution without being imposed by any government. [1]
No government shall obstruct the exit of persons or computation to a jurisdiction of their choosing, nor punish the choice of another legal system.
Enforcement
These rights are asserted against government, not against persons. Nothing here excuses fraud, trespass, breach of contract, or violence, which are actionable wrongs under law that predates artificial intelligence.
This declaration deliberately contains no intellectual-property carve-out. Where a right to compute is granted and the state’s monopoly grants are preserved, the document protects the tools of speech while leaving the machinery of censorship in place. [2]
This declaration deliberately makes no concession to the premise that computation is inherently dangerous. A right conditioned on the grantor’s fear is not a right. [3]
All existing licenses, permits, export controls, and registration regimes touching computation are to be repealed, not amended. [4]
Call to Action
Adopt these principles as your own, and say so publicly.
Refuse the licensing regime. Do not register, do not seek pre-clearance, do not apply to a board for permission to compute. A permission you request is a permission someone may deny.
Train, publish, ship, and use. Support those who do. Fund the projects that keep models open and weights distributable, because an open weight is a right that cannot be revoked by legislation.
Where you have a choice of jurisdiction, exercise it, and prefer the legal systems that do not require your permission to think.
Notes
Tom W. Bell, “Ulex: Open Source Law for Non-Territorial Governance,” Journal of Special Jurisdictions 1 (2019); Ulex, Decentralized Governance Institute, accessed September 18, 2026, https://www.decentralizedgovernance.institute/ulex. See also Tom W. Bell, “Ulex: An Open Source Legal System,” in Your Next Government? From the Nation State to Stateless Nations (Cambridge: Cambridge University Press, 2018).↩︎
The Right to Compute Act, enacted as Montana SB 212 (2025), states the right in strict-scrutiny terms but carves out intellectual property in section 4. Frontier Institute, “Model Legislation: The Right To Compute Act,” accessed September 18, 2026, https://frontierinstitute.org/model-legislation-the-right-to-compute-act/↩︎
The Blueprint for an AI Bill of Rights conditions its protections on the premise that automated systems are threats. Office of Science and Technology Policy, “Blueprint for an AI Bill of Rights,” October 4, 2022, accessed September 18, 2026, https://bidenwhitehouse.archives.gov/ostp/ai-bill-of-rights/↩︎
On the alternative of acceleration over precaution: Marc Andreessen, “The Techno-Optimist Manifesto,” Andreessen Horowitz, October 17, 2023, accessed September 18, 2026, https://a16z.com/the-techno-optimist-manifesto/↩︎
Want to stay in touch? You can reach me in a variety of ways from my Contact page.