New AI Terminal Intelligence Grading Standard Released

On May 8, 2026, China’s Ministry of Industry and Information Technology (MIIT), the State Administration for Market Regulation (SAMR), and two other departments jointly issued the Intelligence Grading Standard for Artificial Intelligence Terminals (GB/Z 177—2026), a national guideline set to take effect on May 13, 2026. The standard introduces the first formal classification framework for AI capabilities in smart building and service terminals—including Smart Lighting controllers, self-service kiosks (Kiosk Tech), and guestroom automation systems—marking a pivotal step toward harmonizing technical claims, certification requirements, and export compliance across high-growth segments of intelligent hardware.

Event Overview

On May 8, 2026, MIIT, SAMR, the Standardization Administration of China (SAC), and the China National Certification and Accreditation Administration (CNCA) jointly released GB/Z 177—2026. The standard defines five intelligence levels (L1–L5) based on verifiable criteria: local inference capability, multimodal response performance, and autonomous decision-making scope. It explicitly includes Smart Lighting controllers, Kiosk interactive terminals, and hotel Guestroom Automation central control panels within its assessment scope. For export products labeled with ‘AI functionality’, mandatory grading certification by CNAS-accredited laboratories is now required prior to market release.

Industries Affected

Direct trading enterprises: Exporters marketing AI-branded terminals to overseas markets—including EU, ASEAN, and Middle Eastern buyers—now face new pre-shipment compliance obligations. Claims such as ‘AI-powered lighting’ or ‘intelligent kiosk’ will trigger verification against L1–L5 benchmarks; unverified labeling may result in customs rejection, post-market recalls, or reputational exposure under evolving global AI transparency regulations.

Raw material procurement enterprises: Suppliers of edge AI chips (e.g., NPU-enabled SoCs), multimodal sensors (e.g., fused audio-visual modules), and low-latency communication components must now align technical specifications with L3+ readiness thresholds—particularly for local inference latency (<100ms) and on-device model update support. Procurement contracts may increasingly reference conformance to GB/Z 177–2026 Annex B test protocols.

Manufacturing enterprises: OEM/ODM producers of Smart Lighting systems, public-access kiosks, and integrated guestroom automation platforms must revise firmware architecture, UI logic, and embedded AI model deployment strategies to meet level-specific validation. For example, L4 certification requires demonstrable fallback autonomy during cloud disconnection—a shift from current cloud-dependent designs.

Supply chain service enterprises: Third-party testing labs, certification consultants, and logistics compliance coordinators are seeing rising demand for GB/Z 177–2026 gap assessments, pre-audit staging, and CNAS-aligned test report generation. Lead times for full-cycle grading certification are currently averaging 22–28 working days, per preliminary data from three accredited labs.

Key Focus Areas and Recommended Actions

Verify current product labeling against L1–L5 definitions

Manufacturers and exporters should conduct an internal audit of all public-facing materials—including datasheets, e-commerce listings, and sales brochures—to identify unsubstantiated ‘AI’ claims. Any reference to AI functionality must map to at least one validated capability under the corresponding level (e.g., L2 requires real-time voice command parsing without cloud round-trip).

Engage CNAS-accredited labs early for baseline testing

Given limited lab capacity and upcoming deadlines for Q3 2026 shipments, firms should initiate pre-certification consultations no later than June 2026. Priority testing should cover local inference throughput, multimodal fusion accuracy (e.g., gesture + voice co-detection), and fail-safe behavior logs—core inputs for L3+ evaluation.

Update R&D roadmaps to support modular intelligence grading

Design teams should adopt a tiered AI architecture: base firmware supporting L1–L2 out-of-the-box, with optional L3–L5 upgrades via certified secure OTA modules. This enables flexible positioning across markets—e.g., L2 for domestic budget deployments, L4 for premium international hospitality contracts.

Editorial Perspective / Industry Observation

Observably, GB/Z 177—2026 does not prescribe technology lock-in (e.g., specific chipsets or frameworks) but enforces outcome-based verification—a notable departure from earlier vertical standards. Analysis shows this approach incentivizes architectural innovation over vendor dependency, especially in edge-AI middleware development. From an industry perspective, the inclusion of Guestroom Automation reflects growing recognition of hospitality tech as a frontline AI integration domain—not merely consumer electronics. Current more critical concern lies in interoperability: while the standard grades individual terminals, it does not address cross-device coordination (e.g., lighting + HVAC + voice assistant handoff), leaving system-level AI claims outside its scope.

Conclusion

This standard signals a maturing phase in China’s AI governance—shifting from principle-based guidance to granular, testable, and trade-relevant technical discipline. It does not raise absolute barriers, but raises the bar for credibility: AI is no longer a marketing adjective, but a graded engineering attribute. For global stakeholders, GB/Z 177—2026 serves less as a restriction and more as a calibration point—helping distinguish substantiated intelligence from speculative capability.

Source Attribution

Official documents published on the websites of the Ministry of Industry and Information Technology (www.miit.gov.cn), the State Administration for Market Regulation (www.samr.gov.cn), and the Standardization Administration of China (www.sac.gov.cn). Full text of GB/Z 177—2026 available for public consultation until June 30, 2026. Implementation details—including approved CNAS labs list and test protocol annexes—are pending official release and remain under active monitoring.