Hello, I understand how you feel.

This may be the sentence that many consumers least want to hear in 2026.

You just want to apply for a refund, but the AI customer service keeps going around in circles. You've already said several times that you want to speak to a human agent, but the system keeps recommending similar answers.

When you finally reach a human agent, the first thing they say is, 'Hello, please describe the problem you're encountering again.' All the communication from the previous half-hour is instantly wiped out.

The good news is that someone is finally addressing this situation.

01 How Hard Is It to Transfer to a Human Agent with AI Customer Service?

Data from the China Consumers Association shows that in the first half of 2026, consumer associations across the country received a total of 985,928 consumer complaints, with after-sales service problems accounting for the highest proportion at 26.79%.

The issues reported by consumers mainly focus on three aspects: false promises by AI customer service, difficulty in accessing human agents, and inaccurate information generated.

Some platforms use methods such as circular responses and multi-level redirects to delay or even prevent consumers from accessing human services.

For complex issues such as returns, refunds, and liability determination, AI customer service cannot handle them in a timely manner, instead increasing the cost of consumers' rights protection.

Even more troublesome is when AI promises don't hold up. Some consumers made transaction decisions based on clear promises from AI customer service regarding fee structures and after-sales policies, only to have the operator later refuse to fulfill them, claiming that AI responses do not represent the company's stance and that algorithm-generated content lacks legal validity.

On third-party consumer service platforms, there are over 4,700 complaints with 'AI customer service' as the keyword.

Posts on social media platforms about how to bypass endless AI customer service to reach a human agent, or complaints about AI customer service repeating irrelevant responses, are abundant.

02 What Exactly Does the New National Standard Stipulate?

On September 1, 2026, China's first national standard focusing on the collaborative mechanism between human and AI customer service, (GB/T 47746—2026), was officially implemented.

The new national standard sets out several mandatory requirements:

The access to human agents must be clear and prominent, not hidden behind multiple layers. When switching between human and AI customer service, a convenient access point must be provided, and the transition must be smooth.

Customer information must be synchronized during human-AI switches, avoiding repeated inquiries. Problems you've already discussed with the AI should not require you to repeat them after transferring to a human agent.

Companies are responsible for the responses provided by AI customer service. They can no longer shirk their obligations by claiming that AI does not represent the company's stance. The new national standard clarifies that AI customer service responses are part of the company's services, and contracts must be honored.

Intelligent customer service should prioritize standardized issues, while human customer service should handle complex and safety-related issues. Simple questions should be handled by AI, while complex disputes should be promptly transferred to human agents.

03 Why Is It So Hard to Transfer to a Human Agent?

The root cause of the difficulty in transferring to a human agent lies not in technology but in cost-cutting.

Wu Yansong, Dean of Caibo Smart Governance Research Institute, who participated in drafting the new national standard, pointed out that many companies, in an attempt to simply save on labor costs, overuse AI customer service, leading to prominent issues such as difficulty in transferring to human agents, irrelevant responses, and uneven service quality.

Some platforms fold the access to human customer service into the last level of the menu, while others have AI repeatedly push the same preset responses, turning the transfer to a human agent into a test of patience.

When consumers input 'human agent' or 'transfer to human agent,' the AI interprets it as a need for 'better assistance' and continues to recommend robot responses.

04 The New National Standard Is Recommendatory

You Still Need to Be Vigilant

It's important to note that the new national standard is a recommendatory national standard and does not directly impose administrative penalties. For this standard to truly take effect, joint efforts from regulators and consumers are needed.

For consumers, there are several practical approaches:

Clearly express your demands. Don't beat around the bush with the AI; clearly state the problem you want to solve in your first sentence. If the AI keeps giving irrelevant responses, directly input 'transfer to human agent' and persist. After the implementation of the new national standard, platforms that continue to obstruct the transfer to human agents are violating regulations.

Retain communication records. Take screenshots of promises made by AI customer service. The new national standard clarifies that companies are responsible for AI responses, and these screenshots serve as evidence for your subsequent rights protection.

Vote with your feet. Give positive feedback when the experience is good, and keep records when it's poor. When companies realize that service quality truly affects user retention, standards will naturally move from the walls into practice.

Ultimately, this new national standard sends a clear signal.

Technology can advance rapidly, but service should not leave people behind. The value of AI customer service lies in handling highly repetitive standardized issues, such as tracking shipments, changing addresses, and return processes.

However, when consumers face complex issues like refund disputes and liability determination, what they need is a human, not a machine.

Next time you're trapped by AI customer service, try saying 'transfer to human agent' directly.

#AICustomerService