Kazakhstan wrote its answers into a dedicated statute before its courts ever met the question. What Kazakhstan does not yet have is practice: no court has tested any of it.
The anchor statute. Definitions including a statutory deepfake concept ("synthetic results"); principles; a three-level risk and autonomy classification with restricted high-autonomy systems; prohibited practices (manipulation, exploitation of vulnerability, social scoring subject to statutory exceptions, emotion recognition without consent); owner and possessor duties with continuous risk management and a duty to suspend; user rights to explanation and to refuse AI interaction where not mandatory; the national AI platform; Article 21 labeling of synthetic content; Article 23 copyright rules for AI. Amending Law No. 326-VIII will adjust several provisions from about 24 August 2026.
Source: adilet.zan.kz/rus/docs/Z2500000230Amends eleven laws to fit the AI statute. Personal Data Law: a new Article 19-1 restricting purely automated decisions that create or end rights, with an objection right; a ban on non-purposive scraping of public sources to build databases; restrictions on biometric collection in public places; consent withdrawal. Consumer Protection Law: sellers must disclose AI components in goods and services. Mass Media Law: AI-produced media content may be distributed only with AI-Law-compliant disclosure.
Source: adilet.zan.kz/rus/docs/Z2500000231Inserts KoAP Article 641-1: fines for failing to inform users about potentially misleading synthetic AI outputs, and for risk-management failures of high-risk systems where the failure caused harm and no criminal offense is made out. Fines scale from 15 to 100 monthly calculation indices by actor size; repeat offenses within a year reach 200 with suspension or prohibition of the AI system. Article 692-3 makes the AI ministry the adjudicator.
Source: adilet.zan.kz/rus/docs/Z2500000232Framework code for the digital sphere: digital government, data governance, digital objects, biometric authentication, cybersecurity, digital-fraud combat. Complements rather than supersedes the AI Law, which remains the special statute; the Code treats AI systems within its "algorithmic system" concept and ties AI-system quality audits to the AI Law's training-data legality and prohibited-functionality checks.
Source: adilet.zan.kz/rus/docs/K2600000255The operative data-protection statute: consent, purpose limitation, security duties. Since 18 January 2026 it carries the AI-era additions from Law No. 231-VIII, including the automated-decision rules and the scraping ban. A further amendment package of 24 June 2026 adds controller and breach registries, mostly not yet in force.
Source: adilet.zan.kz/rus/docs/Z1300000094Governs information systems and e-government, hosts the classification machinery the AI statute reuses, and now routes confidential electronic resources to the national AI platform operator only in anonymized form. Not repealed by the Digital Code.
Source: adilet.zan.kz/rus/docs/Z1500000418The platform and influencer regime: advertising labels, content moderation including Kazakh-language moderation, fake-account rules, fines for spreading false information. Carries no AI-specific provision itself; AI-content labeling arrived through the AI Law and the Mass Media Law instead. Included as the adjacent content infrastructure an AI dispute would also meet.
Source: adilet.zan.kz/rus/docs/Z2300000018Reorganized the digital ministry into the Ministry of Artificial Intelligence and Digital Development, the "authorized body" that supervises the AI Law and adjudicates its administrative cases. Institutional: binds state organs, and stands the regulator up.
Source: akorda.kz decree page; implementing resolution adilet.zan.kz/rus/docs/P2500000846The policy blueprint behind the statute: goals, target indicators and an action plan for compute (supercomputing, the national platform), data, talent and sector adoption through 2029, with annual ministerial reporting. Also the current home of the state's AI ethics principles; a standalone ethics code was proposed but has not been adopted.
Source: adilet.zan.kz/rus/docs/P2400000592The schedule of subordinate acts implementing the AI package: documentation lists, classification criteria, machine-readable-form rules, data-management requirements, and the national AI platform operator designation. Organizational; no private-party duties.
Source: adilet.zan.kz/rus/docs/R2600000002Declares 2026 the Year of Digitalization and Artificial Intelligence: a mobilization instrument that frames the year's legislative and administrative push, without operative duties.
Source: akorda.kz decree pageThe AIFC regulator's survey of generative AI use across authorised firms, expressly non-binding: existing rules plus the FinTech Lab sandbox govern for now, with a re-survey in 2026 before any rule changes. As of 20 July 2026 the AIFC has no AI-specific binding act.
Source: aifc.kz newsAligns the AI Law's vocabulary with the Digital Code (digital objects, cybersecurity), and adds data-controller and breach registries to the Personal Data Law. Adopted but in vacatio legis at this dossier's date; its AI-Law amendments were not yet in force on 20 July 2026.
Source: adilet.zan.kz/rus/docs/Z2600000326Founding member; signed Shanghai, 16 Jul 2026 by the Deputy PM and AI Minister. Official source: China State Council release naming Kazakhstan among the 29; acknowledgment via the AI ministry's press service.
Signatory; 11 Feb 2025. Official source: Elysee official signatory list, entry 31.
Adopted 23 Nov 2021; national readiness assessment launched 16 Jan 2026. Official source: unesco.org announcement of the Kazakhstan RAM launch.
Adopted by consensus 22 Sep 2024; applies to Kazakhstan as a UN member. Official source: A/RES/79/1 annex, un.org.
Consensus resolutions, co-sponsored
Summit-level political commitments
Adopted by the Supreme Eurasian Economic Council, Astana, 29 May 2026, on Kazakhstan's initiative. Official source: Eurasian Economic Commission releases.
Non-binding model legislation
Axes: specificity (x, 0-100) is how squarely the instrument is written for AI; force (y, 0-100) is how hard it binds. Placements are editorial judgments for the comparison plot; the tier column is the authoritative classification.
Tap a marker
The binding and prescriptive corner is the crowded one: the CAC-led measures put force and detail together, service by service. The sector guidance sits just left of the binding line, nonbinding in form but treated as obligatory. Dashed markers are drafts; the National AI Law would consolidate the whole stack into one statute.
Inclusion test: passed. Binding AI-specific operator duties are in force (since 18 January 2026) and observable platforms exist. This section records the duty grid now; per-platform compliance evidence (dated captures) is scheduled for collection at the website build, and platforms are honestly marked unassessed until then.
Distributed AI-generated content that imitates people or events carries machine-readable marking plus a warning the user can actually perceive; the duty sits on the AI system's owner or possessor
No manipulative or subliminal techniques, no exploitation of vulnerability, no social scoring outside statutory exceptions, no emotion recognition without consent, no biometric categorisation for discrimination
Documented risk management for high-risk systems, security measures, user support, and a working ability to suspend or stop the system
Training corpora exclude works whose author or rightsholder expressed a machine-readable prohibition
Users are told when goods, services or interactions are AI-produced, can obtain an explanation of an AI decision, and can refuse AI interaction where the law does not make it mandatory
Mass-media output created with AI is distributed only with AI-Law-compliant disclosure
Evidence capture pending; assess Art. 21 marking of generated media and user-facing AI notices
Evidence capture pending; same criteria
Evidence capture pending; same criteria, plus Russian-language deployment surface
Evidence capture pending; consumer-disclosure duty under Art. 27(6) also applies
Evidence capture pending; synthetic-content distribution surface
Four matters qualify under the inclusion test (AI materially at issue as subject, tool or evidence, with an identifiable procedural reality). Kazakhstan publishes few case numbers; none of the matters below has a published number, and per the fabrication screen none is invented. Context figure, from the interior ministry via press: 59 people were held administratively liable for false or illegal content, a category that officials state expressly covers AI-generated material, in the first five months of 2026.
No voice-cloning civil suit, no copyright litigation over an AI work, and no data-authority enforcement action against an AI system were found.
Kazakhstan wrote its AI-IP answers into a dedicated statute: prompts recognized as works, a machine-readable opt-out for training, and a synthetic-content labeling duty already backed by administrative sanctions.
The AI-IP Index reads a country on two questions that usually get folded together. Accommodation asks how far the law will go to protect intellectual property that a machine helped produce. Institution asks how much administrative machinery sits behind that law: its guidance, its international submissions, its disclosure duties.
A country can be generous on paper and have built almost nothing, or cautious and highly organised, so the two are scored apart and reported as a pair. Each rests on three questions, scored 0, 0.5 or 1.
How far the law will protect intellectual property that a machine helped make.
How much administrative machinery sits behind that law.
The scale runs the same way on both axes: 0 is the open or developed end, 1 is the restrictive or absent one, and 0.5 sits between for anything conditional or untested. Read as a pair, the two scores show whether AI-assisted work can be protected here, and whether the office has said so in writing.
Within each axis the three questions carry equal weight. That is a deliberate choice, not an oversight. An AI developer might reasonably value freedom to train above patent inventorship, but weighting is where indices lose their credibility, so the score stays a flat mean and the choice is stated in the open. Each cell also cites a statute section, a case or an office page, so any single score can be challenged without disputing the rest.
Two positions are worth naming. A country can be open on the law but thin on machinery, an open-but-undeveloped stance that bets on ambiguity. Or it can be restrictive on rights yet run an active office, a restrictive-but-developed stance that reflects a settled policy choice rather than a gap. The pair of scores tells those two apart where a single ranking would blur them.
A rubric is only as good as the results it produces on cases where the answer is already known. These three jurisdictions sit outside this series and are scored with the same six questions, as a check. Their law is the best documented anywhere, so if the scale puts them where the law actually sits, the scale is doing its job. They are reference points, not editions.
Firm judicial settlement at both poles and the weakest statutory position on training. The courts have closed the door on AI inventors and AI authors, while everything on training rides on unresolved fair-use litigation. There is no per se duty to declare AI use in a filing, only a candour duty that bites where the use is material.
Thaler v. Vidal, 43 F.4th 1207 (Fed. Cir. 2022); Thaler v. Perlmutter (D.C. Cir. 2025), cert. denied 2 Mar 2026; Bartz v. Anthropic (N.D. Cal. 2025); USPTO revised inventorship guidance, 90 Fed. Reg. 54636 (28 Nov 2025); 89 Fed. Reg. 58128 (17 Jul 2024); 89 Fed. Reg. 25609 (11 Apr 2024).
The most permissive statutory position on training and the strongest transparency duty in force, sitting over an authorship question no court has answered. Read the label carefully: training and transparency run through EU law, but patents run through the EPO under the European Patent Convention, which is not an EU institution and covers 39 states.
Directive (EU) 2019/790, arts. 3 and 4; Regulation (EU) 2024/1689, art. 53(1)(c) and (d); EPO J 8/20 (2021); EPO Guidelines G-II 3.3.1 (2026 edition); Infopaq C-5/08; Painer C-145/10.
Liberal on patents since the Supreme Court abandoned the Aerotel test in February 2026, conservative on copyright, and currently thin on AI-specific machinery after the dedicated AI examination guidance was withdrawn. Section 9(3) nominally protects a work with no human author for 50 years, the widest AI-output right on paper, but no court has ever applied it to generative AI.
CDPA 1988 ss. 9(3), 12(7), 29A, 178; Thaler v Comptroller-General [2023] UKSC 49; Comptroller-General v Emotional Perception AI [2026] UKSC 3; UKIPO practice notice, 14 Jul 2026; Data (Use and Access) Act 2025 ss. 135 to 137.
The check earned its keep twice. The untested rule held: section 9(3) of the UK Copyright Act reads like the widest AI-output right anywhere, yet because no court has applied it to generative AI it scores as untested rather than open, which is the correct answer and the one a headline reading would miss. The check also exposed a limit worth stating. The Institution axis measures AI-specific machinery, not general office capacity, so a long-established office can score poorly when it has withdrawn its AI guidance and imposes no disclosure duty. Read that axis as what an office has built for AI, nothing wider.
Each row below carries the test we applied and the finding behind the score, with a source. The rationale, not the number, is the point.
Scores verified 20 July 2026, against AIP Index codebook v1.0. A score is only true as of its date.
What we look forA granted patent or a clearly workable examination route, not merely a silent statute.
The findingThe Patent Law confines authorship of industrial-property objects to the natural person whose creative work created them, so the AI-assisted route with a human inventor is open in principle and closed to machines. No examination practice, guidance or grant evidence specific to AI-assisted inventions exists, and no DABUS-type filing has tested Kazakhstan.
What we look forProtection applied where human choices shape the work, and refusal of purely machine output.
The findingA statutory rule is in force: works created using AI systems are protected only where a human made a creative contribution, and prompts that result from human creative activity are recognized as works. Purely machine output has no route to protection. The test is conditional and evidence-demanding, and no court has applied it; a reported first registration of a ChatGPT-assisted graphic novel in December 2025 rests on the applicant's own account in Forbes Kazakhstan and is recorded here as reported, not confirmed.
What we look forA statutory or judicial mining lane. Silence scores against the miner, since the reserved-rights default governs.
The findingThe statute takes a position most jurisdictions have not: training is declared not to presuppose the forms of use covered by authors' moral and economic rights, is expressly excluded from the education and science free-use cases, and is permitted only in the absence of a machine-readable prohibition by the author or rightsholder. On paper that is a permissive opt-out lane; in practice it is six months old, opt-out tooling is nowhere established, and no court or regulator has applied it. The untested-rule discipline holds it at the middle.
What we look forA traceable, substantive submission on the record, not bare membership.
The findingKazakhstan is a WIPO member and cooperates with WIPO on using AI inside examination, but the WIPO Conversation submissions archive lists no Kazakh submission and no substantive session participation was traceable. Membership without a submission earns the middle.
What we look forA standalone AI examination text, not AI handled quietly under general practice.
The findingThe National Institute of Intellectual Property has published no dedicated guidance for examining AI-related filings; examination runs on the general 2018 rules, and the institute's public AI activity concerns using AI as an examination tool. Verified negative, with the caveat that the office's website resists automated crawling.
What we look forA mandatory declaration inside the filing, not a proposal or an informal request.
The findingA mandatory transparency duty is in force and sanctioned: synthetic AI content may be distributed only with machine-readable marking plus a warning the user can perceive, the duty sits on the AI system's owner or possessor, fines scale to 200 monthly indices with suspension or prohibition of the system on repeat, and the AI ministry adjudicates. No patent or copyright filing declaration exists, but the codebook's lever, a binding provenance duty, is present.
Method. Six questions, three per axis, scored 0, 0.5 or 1 against AIP Index codebook v1.0. The three questions in each axis are weighted equally by choice, and the two axes are reported as a pair, never blended into one number. Every score cites a statute, case or office page, carries a verified date, and records its direction of travel apart from the level. Positions are editorial judgements read from primary law and office practice, not official scores. WIPO's public catalogue of IP-office AI initiatives does not score jurisdictions on this basis, so the readings here come straight from the source law. Not legal advice.
Kazakhstan wrote its answers into a dedicated statute before its courts ever met the question. The Law on Artificial Intelligence (No. 230-VIII, in force 18 January 2026) is the first of its kind in Central Asia: output is protected only where a human made a creative contribution, prompts themselves can be works, training a model runs on an express machine-readable opt-out, and AI-generated content may be distributed only with machine-readable marking and a visible warning, with administrative fines and system suspension behind the duty. Around the statute sits a fast-built institutional frame: a Ministry of Artificial Intelligence and Digital Development, a Digital Code in force since July 2026, and an implementation schedule of subordinate acts still rolling out. What Kazakhstan does not yet have is practice: no court has tested any of it.
The site does three things: it maps every instrument that governs AI in Kazakhstan, 21 of them, ordered by how hard each one binds; it reads the same instruments a second time by how squarely each is written for AI; and it records the matters where AI meets the law, with what was searched and found empty stated as plainly as what was found. Every claim traces to a primary source, and absences are recorded rather than papered over.
A membership claim enters this table only from an official list: a declaration annex, a treaty signature table, an organization's own record. Partner and observer statuses never become party status.
Know of a ruling or proceeding in Kazakhstan where AI is central? Send a source link; it will be reviewed before publishing.
Submit a case