On 26 June, the Chinese Institute of Information and Communication announced on 25 June that, under the guidance of the Information and Communication Administration of the Ministry of Industry and Information Technology, the Chinese Institute of Information and Communication, Tell Terminal, United Shenzhen City AI glassesTHE INDUSTRY UNION STUDIED THE DRAFTING OF THE CONVENTION FOR THE SELF-REGULATION OF THE CREDIBLE VISUALIZATION OF AI GLASSES (HEREINAFTER REFERRED TO AS " THE CONVENTION " ) AND RECEIVED A POSITIVE RESPONSE FROM ENTERPRISES UP AND DOWN THE INDUSTRIAL CHAIN。
ON 17 JUNE 2026, THE CONVENTION WAS LAUNCHED AT THE 2026 CONCERTATION FOR THE ECOLOGY OF ARTIFICIAL SMART GLASSES (SHENZHEN). (A) THE EQUIPMENT DIVISION OF THE INFORMATION AND COMMUNICATION ADMINISTRATION OF THE MINISTRY OF INDUSTRY AND INFORMATION TECHNOLOGY, THE CHINA INSTITUTE OF INFORMATION AND COMMUNICATION, THE SHENZHEN CITY OF IAI ALLIANCE FOR THE GLASS INDUSTRY, ANDThunderbird InnovationQuestions, wise cloudsI've got it♪ Billions, Shadows ♪The relevant officials have jointly witnessed the official publication of the Convention。
THE DEPUTY DIRECTOR OF THE TELL TERMINAL LABORATORY OF THE CHINESE INSTITUTE OF INFORMATION AND COMMUNICATION TECHNOLOGY HAS DARED TO STATE THAT, WITH THE DEEP INTEGRATION OF ARTIFICIAL INTELLIGENCE TECHNOLOGY WITH ENHANCED REAL-LIFE TECHNOLOGY, AI GLASSES, AS THE NEXT GENERATION OF INTELLIGENT COMPUTING TERMINALS, ARE ENTERING PUBLIC LIFE AT AN UNPRECEDENTED RATE。The extensive use of audio and video capture modules has put user privacy protection on the testTHE COMPLEXITY AND NON-TRANSPARENTNESS OF AI ALGORITHMS HAVE ALSO RAISED DEEP PUBLIC CONCERNS ABOUT TECHNICAL CONTROL, WHILE THE COLLECTION, TRANSMISSION AND PROCESSING OF LARGE AMOUNTS OF DATA HAVE INCREASED REQUIREMENTS FOR THE GOVERNANCE OF THE FULL CHAIN DATA COMPLIANCE. AI LENSES ARE MOVING FROM “VISIBILITY” TO “UNDERSTAND”, FROM A SINGLE TOOL TO A SMART PARTNER, AND TECHNOLOGICAL PROGRESS MUST BE BASED ON TRUST AND INDUSTRIAL DEVELOPMENT MUST BE PREMISED ON SECURITY。
1AI organizes the main elements as follows:
- AI CONVENTION ON CREDIBLE VISUAL SELF-REGULATION (KEY PROVISIONS)
- Chapter I User privacy protection
- Article 1 Minimum Required Principle
- THE SIGNATORIES UNDERTAKE TO ADHERE STRICTLY TO THE “MINIMUM NECESSARY” PRINCIPLE OF DATA COLLECTION IN THE DESIGN AND DEVELOPMENT OF THE AI GLASSES PRODUCT. ONLY USER DATA NECESSARY TO ACHIEVE THE CORE FUNCTION OF THE PRODUCT ARE COLLECTED, AND USER PERSONAL INFORMATION MAY NOT BE COLLECTED, ABUSED OR DISGUISED OVER-SCOPE. FOR DATA COLLECTION OF SENSITIVE SENSORS, SUCH AS CAMERAS AND MICROPHONES, IT IS IMPORTANT TO PROVIDE CLEAR ACCESS ALERTS AND USER AUTHORIZATION MECHANISMS TO ENSURE USERS ' RIGHT TO KNOW AND CHOOSE。
- Article 2 obligation of transparency
- A commitment to inform users in a clear, understandable and complete manner of the purpose, scope, manner and storage duration of data collection. Privacy policies should avoid lengthy and obscure technical terms and adopt a hierarchical presentation structure to ensure that ordinary users fully understand how their data will be used. Where third-party data-sharing is involved, the recipient ' s identity, the purpose of the sharing and the user ' s exit mechanism must be clearly communicated。
- Article 3 User control
- Products must provide users with easy and effective data management tools, including, but not limited to, one key view of collected data, withdrawal of authorization at any time, bulk deletion of historical data, export of copies of personal data, etc. Upon withdrawal of the authorization by the user, the collection of the relevant data should cease immediately and the data collected should be deleted or anonymously processed in a timely manner as agreed。
- Article IV Biological identity protection
- The collection and use of biometric data, such as iris recognition, facial characteristics, acoustics, etc., require the highest level of security protection. Biometric raw data should, in principle, be processed in local equipment and, if it is necessary to upload the server, it must be encrypted and stored and may not be used for any purpose other than the function of the product。
- Chapter II terminal technology control
- Explanatory nature of the fifth algorithm
- A COMMITMENT TO CONTINUOUSLY IMPROVE THE TRANSPARENCY AND INTERPRETABILITY OF AI ALGORITHMS. FOR KEY DECISION-MAKING (E.G. CONTENT REFERRAL, AUTHENTICATION, RISK TIPS, ETC.) INVOLVING USER INTERESTS, AN UNDERSTANDABLE BASIS FOR DECISION-MAKING SHOULD BE PROVIDED. (C) ESTABLISH AN IMPACT ASSESSMENT MECHANISM FOR ALGORITHMS TO REGULARLY EXAMINE THE RISKS OF BIAS AND DISCRIMINATION THAT MAY RESULT FROM ALGORITHMS AND TAKE TIMELY CORRECTIVE MEASURES。
- Article 6 Localization Priority
- • Promoting “end-of-side intelligence, privacy priorities” technology routes. Priority is given to the completion of data processing and analysis at the end of the equipment and to the reduction of the transfer of raw data to the cloud, subject to the availability of computing power. For data that must be processed at the cloud end, privacy-enhanced techniques such as federal learning and safe multi-computation should be used to achieve “accessibility of data”。
- Article VII security update mechanism
- Establishment of a well-developed mechanism for detecting and repairing security gaps. For known security gaps, patch updates should be published within a reasonable time; for significant security incidents that may affect the privacy of users, they should be reported to the competent authorities within 72 hours of confirmation and notified to affected users within 7 days. The continuous provision of security and up-to-date support over the life cycle of the product must not be less than five years。
- Article 8 open sources and standardization
- ACTIVE PARTICIPATION IN THE DEVELOPMENT OF TECHNICAL STANDARDS AND SAFETY NORMS RELATED TO AI GLASSES. ENCOURAGE THE OPENING OF SECURITY-RELATED TECHNICAL INTERFACES AND TEST TOOLS FOR THIRD-PARTY SECURITY AUDITS, SUBJECT TO COMMERCIAL CONFIDENTIALITY SAFEGUARDS. SUPPORT INDUSTRY TO DEVELOP A UNIFIED SECURITY ASSESSMENT FRAMEWORK AND CERTIFICATION SYSTEM TO FACILITATE THE SHARING OF SECURITY CAPABILITIES。
- Chapter III Full-chain data compliance
- Classification of data under Article 9
- Establish a classification-level management system covering the full life cycle of the data. Discrepancies in collection, storage, use, transmission and removal strategies are developed according to the sensitivity and importance of the data, divided into three levels: general data, key data and core data. The legal and regulatory requirements of the Personal Information Protection Act are strictly adhered to when dealing with personal information。
- Article X Cross-border transmission compliance
- Data exit activities strictly follow the relevant provisions of the Data Exit Security Assessment Scheme. In the case of departure of personal information, legal means such as security assessment, standard contract filing or personal information protection certification are required. Users are clearly informed of the purpose of the cross-border transmission of data, the receiving country/region and the level of local data protection。
- Article 11 Supply chain security management
- Extension of data security requirements to supply chain links. Establishment of security assessment and access mechanisms for partners such as hardware suppliers, software developers and cloud service providers to clarify data protection responsibilities and obligations in contracts. Regular supply chain security audits to ensure the effective implementation of chain-wide data security measures。
- Article 12 Emergency response system
- Establish an emergency response mechanism for data security incidents and develop a plan for a hierarchical response. In the event of a security incident, such as a data leak, tampering or loss, the emergency response process is activated immediately, technical measures are taken to prevent the damage from spreading, reporting to the regulatory authorities as required, and affected users are informed in a timely manner. Response capacity is continuously enhanced through regular emergency exercises。
- Chapter IV Monitoring and publication
- Article 13 Self-examination and audit
- Comprehensive data compliance self-checks are conducted on a regular basis and the introduction of an annual security audit by an independent third party is encouraged。
- Article 14 dynamic amendments
- This Convention is regularly evaluated and revised in the light of the needs of industrial development and technological progress。
- Entry into force of article 15
- THIS CONVENTION SHALL ENTER INTO FORCE ON THE DATE OF ITS PUBLICATION. WE CALL FOR MORE COMPANIES TO JOIN THE CONVENTION, TO SAFEGUARD USER PRIVACY, TO SAFEGUARD TECHNOLOGY CONTROL AND TO MAINTAIN DATA COMPLIANCE, AND TO WORK TOGETHER TO BUILD A CREDIBLE, TRANSPARENT AND SUSTAINABLE AI EYEGLASS INDUSTRY。