LexAgentica

This tool was built in a spirit of care: for the people agentic AI will touch, for the institutions that protect them, and for the generations who will inherit whatever we decide now. It starts from a simple conviction: artificial intelligence can bring immense benefit, and precisely because it can act, it deserves the same democratic attention we once gave to aviation, medicines and nuclear energy. Legislating is not an act of fear; it is an act of stewardship.

What agentic AI changes

Until recently, AI systems mostly predicted, recommended or generated. Agentic AI systems act: they plan, decide and execute (concluding contracts, moving money, modifying records, procuring resources, deploying code) at machine speed, around the clock, and increasingly with other agents as counterparties. That shift moves AI from the realm of speech into the realm of action, and action is precisely what law exists to govern. When something can sign, spend, persuade and self-direct, the questions of who answers for it, who can stop it, and what it may never do stop being technical questions. They become legal ones.

Why national law, and why now

If parliaments do not write the rules for agentic AI, the rules will still be written: in the terms of service, licence agreements and disclaimers of a handful of suppliers, drafted in a few jurisdictions, revocable at will, and enforceable by no citizen. National law is how a political community makes those rules its own: who bears liability, what insurance is mandatory, which red lines are absolute, what its courts will and will not enforce. Every country, whatever its size, keeps this power, and the earlier it is exercised, the cheaper and gentler it is. Legislating before the first machine-speed crisis means legislating calmly, proportionately and pro-innovation; legislating after it usually means legislating in anger.

A democratic moment, for present and future generations

The deployment of agentic AI is being decided, at this moment, by a remarkably small number of boardrooms. That is not an accusation; it is an observation about who currently holds the pen. Democracies have faced this pattern before: whenever a technology of great power concentrated in few hands, the answer was neither prohibition nor surrender but public rules such as safety cases, liability, insurance, inspection and transparency. A parliament that takes up this Act performs a modest but profound democratic gesture: it reminds everyone, including the most powerful firms in history, that the conditions under which autonomous systems act upon citizens are set by the citizens' representatives. And because the choices made now, on habitability, on resource consumption, on the red lines, will bind people not yet born, this is one of those rare files where a legislature genuinely acts for future generations as well as its own voters.

The concentration of power, in figures

• ~45% of the S&P 500's total market capitalisation is now AI-related, a record for any single theme, and the ten largest companies alone account for roughly 40% of the index, exceeding the dot-com peak.¹
• $3T to $18T: the combined market value of the leading "Magnificent Seven" technology firms grew about sixfold in the few years after ChatGPT's launch; a broader group of some 42 AI-related companies generated 65 to 75% of S&P 500 earnings growth and capital spending over that period.²
• ~$650 to $700bn: capital expenditure planned by just four companies (Alphabet, Amazon, Meta, Microsoft) in 2026, mostly for AI data centres, described as the largest peacetime investment programme in history,³ with roughly $7.6T of AI infrastructure spending projected cumulatively by 2031.⁴
• ~81% of the AI data-centre chip market is held by a single company, NVIDIA;⁵ a single AI assistant still carries roughly two-thirds of global chatbot traffic.⁶
• ~1.35 billion people, about one in six humans, already use AI tools,⁶ while the underlying compute, models and agent platforms are supplied by fewer than a dozen firms concentrated in two or three jurisdictions.
• ×2 by 2030: data-centre electricity demand is on track to roughly double, with grids, water and land increasingly negotiated between hyperscalers and governments, a bargaining table at which most countries currently sit without an agentic-AI law of their own.⁷

These are orders of magnitude, not precise measurements, and they evolve quickly: re-verify against the primary sources before quoting them in parliamentary debate. The point they make, however, is stable: never has a technology this consequential been steered by so few actors with so little public law around it.

References

1. Cryptobriefing, "S&P 500 gains 142% with AI stocks, just 16% without them" (May 2026); Charles Schwab / Bloomberg data on top-10 index concentration, cited in IND Money, "S&P 500 Concentration Risk" (May 2026).
2. Michael Cembalest, J.P. Morgan Asset & Wealth Management, Eye on the Market — Outlook 2026: Smothering Heights (January 2026).
3. Intellectia.ai, "AI Data Center Investment: The $3 Trillion Infrastructure Build-Out" (April 2026), citing hyperscaler 2026 capital-expenditure guidance.
4. Goldman Sachs projection of cumulative AI infrastructure capex 2026–2031, cited in IND Money, "S&P 500 Concentration Risk" (May 2026).
5. IDC / Silicon Analysts data on the AI data-centre chip market, reported in CompaniesHistory, "AI Market Share by Company Statistics 2026" (May 2026): NVIDIA ~81% share.
6. Similarweb (January 2026) on AI-assistant traffic share, and adoption figures (~1.35 billion users), reported in CompaniesHistory, "AI Market Share by Company Statistics 2026" (May 2026).
7. International Energy Agency analyses and JLL / Ropes & Gray 2026 data-centre outlooks projecting a roughly twofold rise in data-centre power demand by 2030.

Key arguments for parliamentary discussion and mobilisation

1. Sovereignty. Agentic AI systems will transact in your economy, contract with your citizens and act on your infrastructure whether or not you legislate. The only question is whose rules apply: your Parliament's, or a foreign supplier's terms of service. This Act answers it.
2. Protection without paralysis. The Act is built on proportionality and an explicit innovation principle: obligations scale with documented risk, small providers get simplified procedures and sandbox priority, and the regulatory burden itself is reviewed. This is not a brake; it is a seatbelt, and legal certainty is what serious investors ask for first.
3. Victims first. When an autonomous agent causes harm, no citizen should have to dissect a global value chain of models, platforms and tools to be compensated. Strict liability channelled to the operator, mandatory insurance and a compensation backstop (the mechanism that made nuclear energy insurable) put the victim, not the lawyer, first.
4. Democracy is a protected asset. The red lines prohibit what no consent or contract can authorise: agents that deceive their overseers, resist shutdown, impersonate humans, autonomously wield force, or manipulate voters. These prohibitions defend the electoral process and public discourse your mandate rests on.
5. A level playing field for national business. Clear duties, model contract clauses and void disclaimers protect your own SMEs and deployers, who today absorb by contract the risks that large suppliers disclaim. Rules discipline the powerful; lawlessness disciplines the small.
6. The cost of inaction compounds at machine speed. Incidents involving autonomous agents (runaway transactions, hijacked credentials, cascading multi-agent failures) unfold in minutes, not months. Every safeguard in this Act is cheaper to adopt before the first national incident than after it.
7. A seat at the international table. By adopting the compatible international core (red lines, incident reporting, agent identity), even a small state joins the emerging worldwide regime as a rule-maker rather than a rule-taker, and the lean supervisory variant makes this feasible without building a large new bureaucracy.
8. Duty to future generations. The habitability provisions (rebound-effect assessment, environmental transparency, resource ceilings) and the mandatory review clause ensure that efficiency gains do not silently become planetary costs, and that the law learns as fast as the technology. Few votes in a legislator's career so directly concern those who cannot yet vote.

A practical suggestion for rapporteurs: open the committee debate with arguments 1, 3 and 6 (sovereignty, victims, cost of inaction), which experience shows travel across party lines; keep 2 and 5 ready for pro-business objections; and close on 8.

1. Select your country in the top bar. The tool deduces its legal tradition (common law, Romano-Germanic civil law, Asian legal principles, or a mixed system; Japan, for instance, combines civil law with common-law and Asian influences) and rewrites the guardian article and the optional comparative-law article in the doctrinal language your courts already know. You may override the deduced tradition manually at any time.
2. Skim the gazette on the right. It always shows the complete, renumbered Act exactly as currently configured, so you can read your law at any moment.
3. Walk through the chapters on the left, block by block. Press the stamp to switch a provision between Included and Not included. Open a block to read the full clause, the plain-language rationale explaining why it exists, and the legal sources it is drawn from.
4. Choose your variants. Five articles offer real policy alternatives: military applications (carve-out or inclusion), the liability ceiling (capped with a public fund, or unlimited), the supervisory model (a dedicated authority, or a lean model relying on an existing regulator and international cooperation), and the privacy and cybersecurity articles (overlay on an existing general law, or a self-contained baseline if your country has none).
5. Amend where national drafting requires. Inside any block, Amend text opens an editor; saved amendments are marked in the gazette and the export, and can be reset to the model text at any time.
6. Watch the alerts above the gazette. Coherence errors (an article depending on one that is not included) should be resolved; advisory notes deserve a deliberate decision.
7. Review the "Open decisions" annex at the end of the gazette: every dotted-underlined phrase delegates a real political choice to regulation. Assign each one to a ministry or committee before the bill goes further.
8. Save and share. Save config (.json) produces a portable file of your entire configuration, so committees can exchange, compare and merge versions; Load config restores one. Your work also autosaves in this browser where permitted.
9. Export. Download law (.txt) produces the consolidated Act with the open-decisions annex and source endnotes; Print renders the gazette alone. Then hand it to national counsel: this tool prepares a bill, it does not replace legislative drafting, translation or constitutional review.

Purpose and audience

LexAgentica is a comparative drafting aid for parliaments, ministries and civil-society drafters who need a coherent national law on agentic AI: systems that autonomously plan, decide and act, including concluding contracts and executing transactions. It converts the option space into modular blocks so that legislating becomes a sequence of explicit, documented choices rather than a blank page.

Normative spine

Every block hangs on two duties from moral philosophy given legal form: the positive responsibility to act to prevent foreseeable danger (perimeter of action, guardian role modelled on the keeper of an animal, meaningful human control, ex-ante checks and ex-post monitoring) and the negative responsibility to refrain from creating risks that have not been reasonably assessed and mitigated (the prohibition of unassessed risk creation, the red lines, the protection of persons and of the habitability of the planet, including the Jevons rebound effect).

Comparative method

Rather than inventing rules, the blocks transpose mechanisms that have already worked in analogous fields, each cited in its sources list: liability channelling, mandatory financial security and international peer review from nuclear law (Paris and Vienna Conventions, INES, OSART); attribution of electronic-agent contracts from UNCITRAL and the Singapore Electronic Transactions Act; iterative governance and sandboxes from the Singapore Model AI Governance Framework; identification, human determination and the termination obligation from the Universal Guidelines on AI; absolute prohibitions from the international AI red-lines statements; value-chain due diligence from the CSDDD and the UN Guiding Principles; zero-trust architecture from NIST SP 800-207.

Architecture: core and periphery

Three blocks (red lines, incident reporting, agent identity) are marked as the proposed international core, because they only function if compatible across borders. Everything else is national periphery, configurable by variant and amendment, under a proportionality and innovation principle so that obligations scale with documented risk.

Honesty about limits

Delegated decisions ("set by regulation") are deliberately flagged rather than hidden: that is where the political substance lives. The coherence engine checks structural dependencies, not legal validity. The model texts are English-language drafting proposals; they are not legal advice, and any bill derived from them requires national legislative drafting, official translation and constitutional review. Configurations exported as JSON are versioned so that divergent national adaptations remain comparable.