On October 26, the Jiangxi Provincial Administration for Market Regulation issued the Jiangxi Provincial Guidelines for Compliance in Protecting Trade Secrets of the Aviation Industry (7 chapters, 55 articles) along with the supporting Practical Handbook for Trade Secret Protection for Enterprises in the Jiangxi Aviation Industry (16 reference templates for systems). The announcement was made public on October 28.
Three Key Actions:
Formally included low-altitude economy sectors such as drones and electric vertical take-off and landing aircraft (eVTOL), as well as dual-use aviation technologies, within the definition of the "aviation industry";
Established a three-tier classification system for core trade secrets: Core Trade Secrets / Important Trade Secrets / General Trade Secrets. For information overlapping between state secrets and trade secrets, the principle of "apply the higher standard" is enforced;
Identified four high-frequency leak scenarios: employees taking technical materials upon resignation, circulation of outsourced drawings, cross-border transfer of test flight data, and use of AI tools. Operational norms were provided for each scenario, and administrative complaints, civil litigation, and criminal reporting channels were established as remedies.
01 Not Punitive, But A Benchmark for Future Assessment
Article 52 of the guidelines states: These guidelines are advisory documents and do not have administrative enforcement power.
To qualify as a trade secret, an item must possess secrecy, value, and confidentiality measures. All three elements are indispensable. The first two are determined by technology, while the third depends solely on what the enterprise has done—whether it defined the scope, classified levels, signed agreements, and maintained records. This serves as the basis for determining whether "reasonable confidentiality measures" were taken.
Therefore, the true function of these guidelines is to translate "reasonable confidentiality measures" into an actionable checklist. It does not punish you, but it acts as the yardstick used later to determine if you fulfilled your "confidentiality" obligations.
We must be clear: Infringement of trade secrets can result in orders to cease behavior, confiscation of illegal gains, and fines ranging from CNY 100,000 to CNY 1 million. In serious cases, fines range from CNY 1 million to CNY 5 million. Article 40 also establishes extraterritorial effect—infringements occurring outside China that disrupt domestic competition will still be addressed.
02 Low-Altitude Economy Explicitly Named for the First Time
In Article 2, when defining the "aviation industry," the guidelines added a clause beyond airframe manufacturing, key components, onboard systems, airworthiness services, and test flight testing: "It simultaneously covers emerging fields such as the drone industry, low-altitude economy sectors including electric vertical take-off and landing aircraft (eVTOL), and dual-use aviation technologies."
Previously, low-altitude economy enterprises often asked: Do we qualify as part of the "aviation industry"? Are our flight control algorithms and test flight data protected? Now it is clear: Not only do they qualify, but the guidelines explicitly list these most sensitive assets as trade secrets: core flight control and navigation algorithms, test flight data and experimental protocols, AI-assisted design model parameters, confidential training data, and 3D printing models and prototypes.
03 Three-Tier Classification and One Non-Negotiable Bottom Line
The guidelines classify trade secrets into three tiers, uniformly labeled as "Core Trade Secrets," "Important Trade Secrets," and "General Trade Secrets":
Core Trade Secrets: Original and strategic information critical to survival and development; leakage causes particularly significant losses. Examples include unique processes for key components, core airframe design schemes, material formulations, and core avionics source code;
Important Trade Secrets: Information affecting core business segments; leakage causes significant losses. Examples include complete R&D data, regional core customers, major project bidding proposals, and AI quality inspection parameters;
General Trade Secrets: Leakage causes general losses. Examples include ordinary suppliers, implemented market plans, and non-confidential systems and parameters.
Each tier requires four elements: confidentiality measures, scope of knowledge, approval authority, and confidentiality period. These must be adjusted up or down based on information disclosure and technological iteration. Article 14 warns against "over-classification": defining too broadly increases management costs, and in the event of a breach, the claim that "everything is a secret" may not be accepted.
04 Four Scenarios, Four Most Vulnerable Points
Scenario 1: People Leave, Materials Go With Them. Non-compete restrictions shall not exceed 2 years. Personnel are limited to senior executives, senior technical staff, and others with confidentiality obligations. Restrictions cannot be generalized to the "entire aviation industry field," nor can confidentiality obligations be used as an excuse to refuse compensation—confidentiality is statutory, while non-compete is contractual; they cannot be conflated. Upon job transfer or departure, rights must be revoked immediately ("leave post, lose access"). For AI-related personnel leaving, all storage media, including backups on personal devices, must be cleared. Boundaries must be clear: Knowledge and skills accumulated by employees through their own capabilities do not constitute trade secrets.
Scenario 2: One Outsourced Drawing Leaks an Entire Chain. Conduct due diligence on partners' confidentiality capabilities before cooperation. Follow the "minimum necessary" principle for providing materials; anonymize where possible. Strictly prohibit providing complete design files or process parameters beyond requirements. When Tier 1 suppliers pass information to Tier 2 or Tier 3 suppliers, written consent must be obtained first, and obligations must be transmitted to the end level. Upon contract termination, return or destroy materials within a specified deadline and sign destruction confirmation letters.
Scenario 3: Test Flight Data and the Cross-Border Last Mile. Such data may simultaneously be trade secrets, state secrets, and subject to export controls. The guidelines require establishing export confidentiality reviews, clarifying review entities, processes, and anonymization standards. Submitting materials to the FAA or EASA requires a "dual assessment" of trade secrets and export controls. Core AI trade secrets are strictly prohibited from unauthorized export. For multi-entity collaborative test flights, joint confidentiality agreements must be signed, with data shared according to classification levels.
Scenario 4: AI Tools, The Most Common Leak Point. The red line is explicit: It is strictly prohibited to input, store, or process confidential information in public AI tools. It is strictly prohibited to connect core algorithms or confidential training data to public platforms or third-party cloud services. Supporting requirements are equally specific: Media containing models and training data must be kept by designated personnel and stored encrypted. Copying or transmitting requires approval from core confidential personnel. Confidential and non-confidential data must be stored separately. Operation logs must be retained for no less than the confidentiality period of the trade secret—this provision is specifically designed to preserve evidence for future legal disputes.
05 Rights Protection: Three Paths, One Evidence Set
Administrative complaints, civil litigation, and criminal reporting run in parallel, with administrative-criminal linkage established. Administrative authorities can order cessation of infringement and impose fines. Civil actions can seek pre-litigation injunctions, preservation measures, and punitive damages. Criminal reports can include attached civil lawsuits.
Success depends on evidence. Beyond the three categories of ownership, infringement, and loss, the guidelines specifically add a set for AI: R&D records and proof of ownership for algorithms and model parameters, access logs showing the infringer's contact, and comparison reports between the infringing model and the rights holder's model. To address "difficulty in proving," the guidelines repeatedly recommend notarized preservation, third-party attestation, and fixation of electronic data. This is not formalism; it transforms daily management actions into future evidence.
How to Avoid Risks and Respond?
1. Create Lists: Sort technical and operational information based on "who generates, who identifies, who is responsible," and assign three-tier classifications to each item;
2. Sign Agreements: Execute confidentiality agreements for personnel handling secrets, non-compete agreements for key positions (not exceeding 2 years), and confidentiality clauses for outsourcing and partners, incorporating principles like "minimum necessary" and "written consent required for multi-level transmission";
3. Maintain Ledgers: Keep four ledgers for confidential media, confidential equipment, confidential personnel, and AI models/training data, ensuring consistency between records and physical items;
4. Preserve Traces: Retain classification approval forms, circulation registrations, operation logs, destruction confirmations, and resignation clearance confirmations.
Article 53 also categorizes requirements by scale: Large enterprises should build dedicated systems; medium-sized enterprises should focus on personnel, media, and partners; small enterprises must at least achieve four basics: identification and classification, signing confidentiality agreements, maintaining media ledgers, and including confidentiality clauses with partners.
Conclusion
The signal from these guidelines lies not in their strictness, but in bringing both the low-altitude economy and AI within the scope of trade secret protection.
The core assets of drone and eVTOL enterprises are inherently "data + algorithms + test flight records," not factories and equipment. Previously, they wavered between patents, copyrights, and trade secrets. Now, there is a clear path: classification, tiering, record-keeping, and traceability.