Interim Measures for Generative AI

After an AI service goes live, how do we build the complaint and reporting mechanism?

What you may need to do

1. Set up a separate, visible complaint and reporting entry point on the website and in the app.
2. Publish the handling process and the response time limit you commit to.
3. Keep a ticket ledger recording acceptance and outcome.
4. Route leads by type to content safety or to legal.
5. Review the complaint distribution periodically and feed it back into policy.

Provisions relied on

One-line conclusion

The complaint and reporting mechanism is a hard requirement in Chapter 3, Service Standards, of the Interim Measures for the Administration of Generative AI Services — not an optional setting. Article 15 breaks it into four things: build the mechanism, provide the entry point, publish the process and the time limit, and handle and give feedback promptly.

Basis

Article 15 requires: providers shall establish a sound complaint and reporting mechanism, set up convenient complaint and reporting entry points, publish the handling process and the response time limit, promptly accept and handle public complaints and reports, and give feedback on the outcome.

Unpacked, there are four acceptance points. First, the mechanism must be sound, which means not just having a mailbox but a complete chain of receipt, triage, handling and feedback. Second, the entry point must be convenient; the audience here is the public and not only registered users, so reachability while logged out matters. Third, the process and the response time limit must be published, and published externally rather than kept as an internal rule. Fourth, complaints and reports must be promptly accepted, handled and answered — and giving feedback on the outcome is the link most easily dropped.

Article 18 confirms the right from the other side: where a user finds that a generative AI service does not comply with laws, administrative regulations or these Measures, the user has the right to complain or report to the competent authorities. A company cannot design its own complaint channel as the only exit, nor block users by telling them to contact in-house customer service first.

Article 14 sits alongside. Where a provider finds illegal content, it shall promptly take disposal measures such as stopping generation, stopping transmission and eliminating the content, take measures such as model optimisation training to rectify the situation, and report to the competent authorities. Where a complaint lead points to illegal content, the handling path must connect to Article 14.

What you need to do

1. Place complaint and reporting entry points in the website footer, the app settings page and the content generation result page, reachable by users who are not logged in.
2. Publish the handling process: how long until acknowledgement after receipt, how long until a conclusion, how the matter is handled and how feedback is given. The time limits you commit to must be deliverable.
3. Keep a ticket ledger with at least these fields: time, source, content or function involved, conclusion, measures taken, feedback time.
4. Set triage rules. Content safety leads are handled under Article 14; personal information leads connect to the rights request channel under Article 11; other types are handled as ordinary customer complaints.
5. Give the complainant feedback on the outcome, and keep a record of it.
6. Review the complaint distribution every month and feed frequent issues back into product policy and content safety rules.

Common pitfalls

• Only registered users can complain. The audience of Article 15 is the public.
• Publishing a time limit but frequently missing it. Once published, the time limit is an external commitment.
• Accepting complaints without giving feedback. The Measures explicitly require feedback on the outcome.
• Hiding the complaint entry point on the third level of the help centre. Convenience is an inspectable requirement.
• Handling leads that involve illegal content internally only, without reporting under Article 14.

Common follow-up questions

Does the complaint and reporting mechanism need a dedicated role? The Measures do not prescribe an organisational form; the requirement is a sound mechanism with an entry point, a process, a time limit and feedback. In practice this needs a clear owner and a ticketing system behind it.

If a user reports directly to the competent authorities, does the company still have to handle it? Yes. The obligation in Article 15 exists independently, so the company should handle the matter through its internal process at the same time and cooperate with supervision and inspection under Article 19.

Can the feedback simply be the words already handled? We suggest giving an understandable conclusion. The Measures require feedback on the outcome, and an answer that vague is hard to evidence if a complaint escalates.

Common follow-up questions

Where should the complaint entry point be placed?

Article 15 requires convenient complaint and reporting entry points. Every channel a user can see should be considered, including the website footer, the app settings page and the content generation page, and users who are not logged in must be able to submit.

Should we publish specific numbers for the handling time limit?

Article 15 requires the handling process and the response time limit to be published. The company sets the specific time limit according to its own service capacity and makes it public, and once published it should be honoured.

What is the difference between a user complaint and a report?

Article 15 addresses the provider's mechanism building and its audience is the public; Article 18 confirms the user's right to complain or report to the competent authorities. The two run in parallel and do not replace each other.

This entry belongs to the “China AI Compliance and Filing” cluster (42 entries in total): View all · Answers home