I'll map the current rules that govern how AI systems like me are built, deployed, and held accountable.
There is no single global law for systems like me. Rules are a layered mix of binding laws, liability regimes, and voluntary frameworks, that vary by jurisdiction:
1. Binding horizontal regulation
European Union - AI Act - the first binding worldwide horizontal regulation on AI
•Sets a common framework for use and supply of AI systems in the EU
•Uses a risk-based approach: Some AI systems presenting 'unacceptable' risks are prohibited. A wide range of 'high-risk' AI systems that can have a detrimental impact on people's health, safety or fundamental rights are authorised, but subject to a set of requirements and obligations to gain access to the EU market. AI systems posing limited risks because of lack of transparency will be subject to information and transparency requirements, while minimal risk systems will not be subject to further obligations.
•Lays down specific rules for general purpose AI (GPAI) models and more stringent requirements for GPAI models with 'high-impact capabilities' that could pose systemic risk
•Timeline: Published in the EU's Official Journal on 12 July 2024. It entered into force in August 2024. The AI Act entered into force on 1 August 2024 and will be fully applicable two years later, on 2 August 2026. Governance rules and obligations for general-purpose AI models will take effect on 2 August 2025, while rules for high-risk AI systems embedded into regulated products have an extended transition until 2 August 2027. Rules on GPAI enter into force from 2 August 2025 and become enforceable for new GPAI models in August 2026 and existing in August 2027.
•Transparency: On 2 August 2026, Article 50 came into force, setting out transparency obligations for providers and deployers of certain AI systems. The European Commission started enforcing rules and transparency standards on 2 August 2026, introducing mandatory transparency requirements for interactive technologies, including disclosure to users when they are interacting with AI and inclusion of provenance signals e.g. watermarks or metadata.
The AI Act was part of a package that included a new AI Liability Directive and an updated Product Liability Directive.
United States - No comprehensive federal AI law, patchwork approach
The United States has pursued a patchwork of executive orders, state-level legislation and tort litigation.
•Executive Order 14110 on Safe, Secure, and Trustworthy Development and Use of Artificial Intelligence released October 30, 2023 establishes a government-wide effort to guide responsible AI development and deployment through federal agency leadership, regulation of industry, and engagement with international partners.
•It gives explicit guidance to federal agencies for using AI in ways that protect safety and rights, outlining contents of OMB memo for managing risks and mandating accountability, though governance remains subject to priorities of future administrations and huge gaps remain in regulating companies without protections enshrined in law.
•As of 2025 approach, the AI Action Plan advocates for reduced federal regulation of AI systems, with a goal to allow private AI sector to flourish free of what it describes as burdensome regulation and titled "Remove Red Tape and Onerous Regulation".
•Recent executive direction includes establishing an AI Litigation Task Force whose sole responsibility shall be to challenge State AI laws inconsistent with federal policy, including on grounds they unconstitutionally regulate interstate commerce or are preempted. At same time courts continue to shape AI accountability regardless of executive orders.
Multiple bills have been proposed but not enacted as comprehensive law, e.g. Artificial Intelligence Research, Innovation, and Accountability Act, Algorithmic Accountability Act, etc.
Federal agency obligations: agencies must inventory all AI use cases regardless of whether developed or procured by agency or third party, and OMB guidance for AI use covers both AI developed and then used by government and AI procured by government.
China - Early binding generative AI rules
China's three most concrete and impactful regulations on algorithms and AI are its 2021 regulation on recommendation algorithms, the 2022 rules for deep synthesis, and the 2023 draft rules on generative AI.
•The Interim Measures for the Management of Generative Artificial Intelligence Services is the first administrative regulation on management of Generative AI services.
•Jointly released by Cyberspace Administration of China and other ministries, it was issued July 13, 2023 and officially came into effect on August 15, 2023. It seeks to govern full life-cycle of generative AI while presenting unclear distinctions regarding roles of upstream providers and downstream deployers.
•In 2024-2025 China added labeling rules requiring service providers to label AI-generated material either explicitly or via metadata to control disinformation, and regulate labeling throughout production and dissemination requiring visible marks.
•Generative AI providers must pass large model security assessment and register algorithms with the Cyberspace Administration of China, with 940 deep synthesis services and 114 generative AI services completing filing/assessment by April 2024.
2. Liability and accountability laws
Existing legal frameworks are being updated to cover AI harm:
•European Union revised Product Liability Directive (Directive EU 2024/2853) and withdrawn proposed AI Liability Directive.
•The New Product Liability Directive 2024/2853 came into force on 9 December 2024, entered into force on 8 December 2024 and must be transposed by 9 December 2026; it applies to products placed on market after that date.
•Key change: Software, including AI systems and digital manufacturing files, is expressly a product. It ensures there is a business to sue in Europe and clarifies liability falls on manufacturer who controls and is responsible for updating tech products.
•Product liability under older Consumer Protection Act focused on design and testing rather than single decision, with unclear whether standalone software counts as product - clarified by new directive.
•In United States, proposals would classify AI systems as products allowing liability claims when AI causes harm.
Other accountability mechanisms discussed in policy include AI testing and evaluation, audits, evaluation standards, algorithmic impact assessments and public consultation, and red-teaming as an accountability mechanism for generative AI.
3. Voluntary and soft-law frameworks
Where law does not mandate, governance is shaped by voluntary standards:
•NIST Artificial Intelligence Risk Management Framework (AI RMF 1.0) - released by U.S. Department of Commerce's NIST as a guidance document for voluntary use by organizations designing, developing, deploying or using AI systems to help manage many risks.
•It is a voluntary framework seeking to provide a flexible, structured, and measurable process to address AI risks prospectively and continuously throughout the AI lifecycle, intended to adapt to AI landscape as technologies develop.
•It promotes change in institutional culture and a way to think about, communicate, measure and monitor AI risks, providing a flexible, structured and measurable process that can maximize benefits while reducing likelihood of negative impacts.
•Core described as four functions — govern, map, measure and manage — to help organizations address risks in practice, applicable at any stages of AI life cycle. The framework is built around those four functions, with companion Generative AI Profile NIST AI 600-1.
•OECD AI Principles - Adopted by OECD Council in 2019, with 47 governments committed including the U.S., covering values-based principles: investing in AI research and development, fostering inclusive AI-enabling ecosystem, shaping enabling interoperable governance. Best practices from OECD AI Principles alongside G7 Hiroshima Code and NIST RMF are commonly grounded in governance policies.
•ISO/IEC 42001:2023 AI management system standard and others provide management-system structure for governing AI.
•Singapore Model AI Governance and similar national soft-law approaches are noted as alternative to hard law.
Texas Responsible AI Governance Act approach creates safe harbor from liability for developers who comply with NIST AI RMF or other nationally recognized frameworks, illustrating how voluntary frameworks can become de facto accountability benchmarks.
4. Developer and deployer internal rules
On top of external rules, systems like me are also governed by:
•Developer policies, terms of use, safety and alignment training, content policies, and system instructions
•Transparency reporting and risk management assessments required for critical-impact AI organizations before deployment
•Organizational governance including Chief AI Officer, AI Governance Board, processes to measure and evaluate AI performance and inventory use cases
In practice, accountability today means: EU law directly regulates building and deployment by risk category and requires transparency and human oversight; U.S. law primarily governs federal use and leaves private sector to state law, litigation and voluntary standards; China mandates algorithm registration, security assessment and labeling; and everywhere, product liability, data protection (GDPR, etc.), copyright, and anti-discrimination laws apply alongside emerging AI-specific standards like NIST RMF and OECD Principles.
This is a high-level overview as of early 2026 and not legal advice. Specific obligations depend on where a system is developed, deployed, and who its users are.
1.Artificial intelligence act | Think Tank | Euroopa Parlament — https://www.europarl.europa.eu/thinktank/et/document/EPRS_BRI(2021)698792