| Region | Primary bodies | Binding or voluntary |
|---|---|---|
| International baseline | ISO, IEC | Voluntary unless adopted into national law |
| United States | ANSI, UL, RIA, A3 | Voluntary consensus standards, some referenced by OSHA |
| European Union | CEN, CENELEC, EU Machinery Regulation, EU AI Act | Binding for market access via CE marking |
| China | MIIT, SAC, GB national standards | Binding, state-directed, rapidly expanding |
| Japan and South Korea | JISC, METI, KATS | Mixed: some binding safety codes, some voluntary guidance |
Why Robotics Regulation Looks Nothing Like a Single Global System
Ask a robotics engineer shipping hardware into three markets at once which regulatory framework applies, and the honest answer is: several, simultaneously, and they do not always agree with each other. Robotics sits at the intersection of machinery safety, software and AI governance, data protection, and in some cases telecommunications regulation, and no single global body has authority over all of it. What exists instead is a layered landscape: international standards bodies that publish technical baselines nobody is legally required to follow unless a government adopts them, regional regulators that turn some of those baselines into binding law, and national standards bodies that sometimes go well beyond the international baseline entirely.
This matters for anyone building, deploying, or governing robotic systems because compliance in one region does not imply compliance in another. A collaborative robot certified against a European safety standard is not automatically compliant with Chinese national standards, and a US-market industrial robot meeting ANSI and RIA guidance has no formal standing under the EU’s CE marking regime without a separate conformity assessment. This article maps that landscape region by region, focused specifically on who writes the rules, what force those rules carry, and how the pieces fit together.
The International Baseline: ISO and IEC
Two international bodies sit underneath almost every regional robotics regulation in some form: the International Organization for Standardization and the International Electrotechnical Commission.
ISO Technical Committee 299 and Robot Safety
ISO’s Technical Committee 299 is the primary body responsible for robotics-specific standards. Its flagship documents, ISO 10218-1 for robot manufacturers and ISO 10218-2 for system integrators, define baseline safety requirements for industrial robots covering mechanical design, control systems, and integration into a workcell. ISO 10218 was updated in 2025, and much of the world’s national robotics regulation either directly references it or was substantially modeled on it. TC 299 is also actively developing newer standards for categories the original 10218 framework did not anticipate, including ISO 25785-1 for bipedal robot safety and ISO 26058-1 for dynamically stable mobile robots, both still in earlier stages of development as of 2026.
ISO/TS 15066 and Collaborative Robots
ISO/TS 15066 supplements ISO 10218 specifically for collaborative robot applications, defining the force and pressure thresholds that determine safe human-robot proximity work. It is one of the most frequently referenced technical specifications in cobot deployments worldwide, precisely because so few national regulators have written their own equivalent from scratch.
IEC and Electrical, Functional Safety Standards
The IEC’s contribution sits one layer below ISO’s robot-specific standards, covering the electrical and functional safety systems that robots depend on. IEC 61508 for functional safety of electrical and programmable systems, and its machinery-specific derivative IEC 62061, underpin the safety-rated control systems referenced throughout ISO 10218. Neither ISO nor IEC standards carry legal force on their own; they become binding only when a national or regional regulator formally adopts or references them, which is exactly what happens in most of the regions covered below.
The United States: A Voluntary Consensus Model
The US robotics regulatory landscape looks fundamentally different from the EU’s, built around voluntary consensus standards rather than a mandatory pre-market certification regime.
ANSI and the A3 Partnership
The American National Standards Institute does not write robotics standards itself; it accredits the process by which industry bodies develop them. The Association for Advancing Automation, formerly the Robotic Industries Association, is the accredited standards developer for US industrial robot safety, and its flagship document, ANSI/RIA R15.06, was updated and re-released in 2025 to align more closely with the international ISO 10218 baseline. Critically, ANSI standards are voluntary consensus standards; they do not represent legally binding market access requirements the way CE marking does in Europe, though they are frequently referenced in litigation and insurance underwriting as evidence of prevailing industry practice.
UL and Product Safety Certification
UL operates as a nationally recognized testing laboratory that certifies robotic products against its own safety standards, such as UL 3100 for autonomous mobile robots. UL certification is not a legal requirement in most US jurisdictions, but it functions as a practical market-access gate because many insurers, large purchasers, and facility operators require it as a condition of deployment even where no statute demands it.
OSHA’s Indirect Role
The Occupational Safety and Health Administration does not publish robotics-specific standards but enforces workplace safety through its general duty clause, and it frequently treats ANSI/RIA and ISO standards as evidence of the recognized industry standard of care during an investigation. This creates an unusual dynamic where a technically voluntary standard becomes practically mandatory through enforcement precedent.
The European Union: A Dual Regulatory Track
Europe presents the most binding and, as of 2026, the most complex regulatory environment in the world for robotics, because two major regulatory instruments now apply simultaneously to many robotic products.
The EU Machinery Regulation
The EU Machinery Regulation, replacing the older Machinery Directive, governs the physical safety of machinery placed on the EU market, including most industrial and service robots. Conformity requires CE marking, and the regulation carries a hard enforcement date of January 20, 2027, with no grace period, meaning products placed on the market after that date must fully comply with its documentation and safety requirements, including specific provisions addressing machinery with self-evolving or AI-driven behavior.
The EU AI Act’s Overlapping Scope
Where a robot incorporates an AI system that falls under the AI Act’s high-risk categories, both the Machinery Regulation and the AI Act apply, and a manufacturer must assess each component’s classification independently. A robot’s navigation AI might be regulated primarily through the Machinery Regulation’s safety framework, while a separate AI function such as facial recognition integrated into the same robot could trigger the AI Act’s Annex III high-risk obligations. The AI Act’s transparency and high-risk provisions became enforceable in stages through 2026, with conformity assessments, technical documentation, and EU database registration required for in-scope high-risk systems.
CEN and CENELEC
The European Committee for Standardization and the European Committee for Electrotechnical Standardization are the bodies that translate ISO and IEC international standards into harmonized European standards, which manufacturers can use as a presumption of conformity with the underlying EU regulations. This is the formal mechanism by which ISO 10218 effectively becomes a European requirement rather than merely an international recommendation.
Figure: How a single ISO standard becomes binding law in different regions
ISO 10218 originates as a voluntary international standard. In the European Union, CEN and CENELEC formally harmonize it, linking it to the Machinery Regulation so that compliance becomes a precondition for CE marking. In the United States, ANSI accredits the A3-developed R15.06 standard, which closely tracks ISO 10218 but remains voluntary unless referenced in litigation or contractually required by a customer. In China, the same underlying safety principles are absorbed into binding GB national standards issued through MIIT and SAC, with compliance mandatory for market access. The identical technical content ends up carrying three different levels of legal force.
China: State-Directed and Rapidly Expanding
China’s approach to robotics standards differs from both the US voluntary model and the EU’s regulation-plus-harmonized-standards model in that its national standards are issued directly by state ministries and are binding by default once published.
The Ministry of Industry and Information Technology’s Role
MIIT coordinates national industrial policy for robotics and has taken an increasingly active role in standards-setting. In February 2026, MIIT unveiled the Humanoid Robot and Embodied Intelligence Standard System, known as HEIS 2026, developed collaboratively with more than 120 research institutions, manufacturers, and industry end-users. This standard system is notably broader in scope than the equivalent international work at ISO, which as of 2026 is still developing narrower standards like ISO 25785-1 for bipedal robot safety.
GB National Standards and the June 2026 Expansion
China’s GB standards, issued through the Standardization Administration of China, function as the country’s binding national standards, roughly analogous to a mandatory version of what ISO standards are internationally. On June 1, 2026, China put 102 new national standards into effect, a significant portion of which touch robotics, embodied AI, industrial automation, and AI agents directly. This represents one of the largest single expansions of binding robotics-adjacent standards anywhere in the world within a single year.
Prove, Don’t Demo
A distinguishing feature of China’s approach is its emphasis on demonstrated conformance testing rather than self-certification or documentation review alone, reflecting a broader regulatory posture that leans toward direct verification over the more paperwork-driven compliance regimes common in the US and EU.
| Characteristic | United States | European Union | China |
|---|---|---|---|
| Primary legal mechanism | Voluntary consensus standards, indirect OSHA enforcement | CE marking under binding regulation | Binding GB national standards |
| Key robotics-specific document | ANSI/RIA R15.06-2025 | EU Machinery Regulation plus AI Act | HEIS 2026 and GB standards (2026 expansion) |
| Enforcement style | Litigation and insurance-driven | Pre-market conformity assessment | State-directed conformance testing |
| Pace of regulatory change | Incremental, industry-led | Structured, deadline-driven (2026 to 2027) | Rapid, centrally coordinated |
Japan and South Korea: Established Safety Codes, Lighter AI-Specific Rules
Japan and South Korea both have long-established industrial robot safety traditions predating much of the current global regulatory activity, but neither has moved as aggressively as the EU or China on AI-specific robotics rules.
Japan’s METI and JISC Framework
Japan’s Ministry of Economy, Trade and Industry sets robotics-relevant industrial policy, while the Japanese Industrial Standards Committee maintains JIS standards that closely track ISO’s international work, reflecting Japan’s long history as an early adopter and co-developer of international industrial robot safety standards going back decades. Japan’s regulatory posture on AI governance more broadly has favored soft-law guidance and sectoral codes of conduct over the kind of binding, horizontal AI statute the EU has adopted.
South Korea’s KATS and Emerging AI Framework
South Korea’s Korean Agency for Technology and Standards administers KS national standards, which similarly track ISO and IEC baselines for industrial robot safety. South Korea has been moving toward more formal AI governance legislation in recent years, but its robotics-specific regulatory apparatus remains closer to the US model of safety-code enforcement than to the EU’s dual-track regulation and AI Act structure.
Common mistake
Teams frequently assume that meeting ISO 10218 automatically satisfies every regional requirement, since it is the most widely referenced international robot safety standard. In practice, ISO compliance is a necessary starting point in most regions but not a sufficient one: the EU requires separate CE conformity assessment and AI Act classification where applicable, the US market often expects UL certification on top of ANSI/RIA alignment, and China requires compliance with its own binding GB standards regardless of ISO alignment elsewhere. Treat ISO 10218 as a shared technical vocabulary, not a universal compliance passport.
Building a Multi-Region Compliance Map
Organizations shipping robotics products into more than one of these regions benefit from treating regulatory mapping as an ongoing engineering deliverable rather than a one-time legal review.
Start With the Strictest Applicable Regime
Designing to the most demanding applicable requirement, typically the EU’s combined Machinery Regulation and AI Act obligations for AI-enabled robots, tends to produce documentation and testing artifacts that can be adapted downward for less demanding regimes more easily than the reverse.
Track Enforcement Dates Independently by Region
The EU Machinery Regulation’s January 20, 2027 enforcement date, the AI Act’s staged 2026 high-risk deadlines, and China’s rolling GB standard updates do not move in sync, and a compliance calendar built around a single region’s timeline will quietly fall out of date elsewhere.
Assign Ownership for Each Region’s Standards Body Relationships
Larger organizations increasingly assign a named owner to track ISO TC 299 working group activity, CEN and CENELEC harmonization progress, ANSI/A3 committee updates, and China’s MIIT and SAC publication schedule, since these bodies rarely coordinate their release timing with each other.
What worked
One robotics manufacturer selling into North America, the EU, and China maintained a single internal spreadsheet mapping every applicable standard to its region, enforcement status, and next review date, updated quarterly by a designated compliance owner rather than left to whichever engineer happened to notice a new publication. When the EU AI Act’s staged obligations began taking effect through 2026, the team already had a documented mapping showing exactly which of their product lines were affected, avoiding the scramble that caught several competitors who had been tracking EU rules only informally.
Frequently Overlooked Details in the Standards Landscape
- Harmonized standards versus base regulationA harmonized European standard gives a presumption of conformity with the underlying regulation, but it is not itself the legal requirement; the regulation remains the binding instrument even when no harmonized standard yet exists for a given product category.
- Dual classification under the EU AI ActA single robot can contain multiple AI components that each require separate classification, meaning one product may need to satisfy several different obligation tiers simultaneously rather than a single overall rating.
- Voluntary standards used as legal evidenceUS voluntary consensus standards like ANSI/RIA R15.06 are not legally mandatory, but courts and OSHA investigators routinely treat deviation from them as evidence of negligence, which functionally raises their importance well above their nominal voluntary status.
- China’s conformance testing emphasisChina’s regulatory culture favors direct, demonstrated testing over paperwork-based self-certification, which means documentation packages built for EU or US audiences often need a genuinely separate testing and demonstration process for the Chinese market.
- Standards bodies do not synchronize release timingISO, CEN and CENELEC, ANSI and A3, and China’s MIIT and SAC all publish and revise on independent schedules, so a compliance program built around any single region’s calendar will drift out of alignment with the others over time.
Key Takeaways
Key Takeaways
- Robotics regulation is a regional patchwork, not a single global system, and compliance in one region does not automatically transfer to another.
- ISO and IEC set voluntary international technical baselines that individual regions adopt, harmonize, or exceed in different ways.
- The United States relies primarily on voluntary consensus standards through ANSI, A3, and UL, enforced indirectly through OSHA precedent and litigation rather than pre-market certification.
- The European Union runs a binding dual-track system combining the Machinery Regulation and the AI Act, requiring separate conformity assessments and CE marking.
- China issues binding GB national standards directly through state ministries, with a major 102-standard expansion taking effect in June 2026 and a dedicated humanoid robot standard system launched in February 2026.
- Japan and South Korea maintain established, ISO-aligned safety codes but have moved more cautiously than the EU or China on binding AI-specific robotics rules.
- Organizations operating across multiple regions benefit from designing to the strictest applicable regime and tracking each region’s enforcement calendar and standards body activity independently.
Glossary
- ISO 10218
- The core international standard for industrial robot safety, split into a manufacturer-facing part and an integrator-facing part, widely referenced or adopted across most national regulatory regimes.
- CE marking
- The conformity marking required for machinery, including most robots, placed on the European Union market, indicating compliance with applicable EU regulations such as the Machinery Regulation.
- Harmonized standard
- A European standard developed by CEN or CENELEC that, when followed, gives a manufacturer a presumption of conformity with a corresponding EU regulation without proving compliance from first principles.
- GB standard
- A Chinese national standard issued through the Standardization Administration of China, binding by default and covering an increasing share of robotics, embodied AI, and industrial automation requirements.
- Voluntary consensus standard
- A standard, such as those developed by ANSI-accredited bodies in the United States, that carries no automatic legal force but is frequently treated as evidence of the recognized industry standard of care.
FAQs
Is ISO 10218 legally required anywhere in the world?
Not directly on its own; ISO standards carry no automatic legal force. It becomes effectively mandatory where a region formally adopts or harmonizes it, as the European Union does through CEN and CENELEC in connection with the Machinery Regulation, or where national standards closely mirror it, as with US ANSI/RIA R15.06.
What is the difference between the EU Machinery Regulation and the EU AI Act for robots?
The Machinery Regulation governs the physical safety of the machine itself, while the AI Act governs the AI systems the machine may contain, particularly high-risk applications. A single robot can be subject to both simultaneously, requiring separate conformity assessments for each applicable component.
Are US robotics standards legally mandatory?
Generally no. ANSI-accredited standards like A3’s R15.06 are voluntary consensus documents. However, OSHA and courts frequently treat deviation from them as evidence of failing to meet the recognized industry standard of care, giving them significant practical weight despite their voluntary status.
How is China’s approach to robotics standards different from the US and EU?
China issues binding GB national standards directly through state ministries such as MIIT, with compliance mandatory by default rather than voluntary or conditional on market-access certification. China also emphasizes demonstrated conformance testing more heavily than the paperwork-driven compliance regimes common elsewhere.
What happened with robotics standards in China in 2026?
MIIT unveiled the Humanoid Robot and Embodied Intelligence Standard System in February 2026, and on June 1, 2026, China put 102 new national standards into effect, a significant number of which apply directly to robotics, embodied AI, and industrial automation.
Do Japan and South Korea have binding AI-specific robotics regulations like the EU AI Act?
Not to the same degree as of 2026. Both countries maintain established, ISO-aligned industrial robot safety codes, but their approach to AI governance more broadly has leaned toward guidance and sectoral codes rather than a single binding horizontal AI statute comparable to the EU AI Act.
What does CE marking actually certify for a robot?
CE marking indicates that a product complies with applicable EU regulations, which for a robot typically includes the Machinery Regulation and, where relevant, the AI Act for embedded AI functions, based on a conformity assessment the manufacturer must complete before placing the product on the EU market.
What is the single biggest mistake companies make when navigating multiple robotics regulatory regimes at once?
Assuming compliance in one region transfers to another. ISO alignment, CE marking, UL certification, and Chinese GB standard compliance are separate processes with separate documentation, testing, and enforcement mechanisms, and treating any one of them as universally sufficient is the most common and costly planning error.
For the argument that robotics governance has crossed the line into a genuine legal mandate rather than best practice, see our companion piece on why robotics governance is now a legal requirement. Readers interested in how provenance and audit trail data intersect with this regulatory backdrop should see our deep dive on digital provenance in robotics. For how these regional rules connect to the insurance market’s own response to autonomous risk, see our analysis of liability and insurance for autonomous machines. Teams building the safety case that regulators expect should also consult our explainer on functional safety for robots, and organizations running ROS 2 fleets subject to these regional rules may find our guide on ROS 2 in production useful for the engineering side of compliance.
References
- ISO, Technical Committee 299, “Robotics” work program and published standards, 2025 to 2026
- EVS International, “Industrial Robot Safety Standards: ISO 10218, ISO/TS 15066 and CE Marking (2026)”
- EVS International, “China’s 102 Standards on Robotics and AI: Prove, Don’t Demo”
- RobotToday, “China’s Humanoid Robot and Embodied Intelligence Standard System (HEIS 2026)”
- Automate.org, “Global Robotic Standards,” Association for Advancing Automation
- Bird and Bird, “Smart Robots, Dual Regulations: Navigating the AI Act and Machinery Compliance,” 2026
- Nemko Digital, “Guide to the 2027 EU Machinery Regulation”
- European Commission, “AI Act,” Shaping Europe’s Digital Future, Directorate-General for Communications Networks, Content and Technology
- Process and Control Today, “How Ready Are Suppliers for New Robot Safety Standards?”
