Summary in Plain Language
The new regulations on AI dispute adjudication issued by the Supreme People's Court in September 2026 are a real boon for open-source AI developers. As long as you clearly inform all users of the functionality and potential risks of your open-source AI code/models in advance, you won’t be held responsible if someone else uses that code for infringement or illegal activities. However, lawyers have reviewed the user agreements of major open-source AI products in China and found that they all contain outdated disclaimers from the traditional software era, which do not meet the new regulations’ requirements for exemption from liability. This means that the so-called “immunity” currently available is essentially worthless, and failure to comply could result in significant losses.
---
Detailed Analysis
1. Understanding the “Immunity Pass” Granted by the Supreme Court
Many people mistakenly think that since they provide open-source AI for free, they should not be held responsible. In the past, without clear regulations, it was common for downstream users to use open-source code for malicious purposes, and the developers were often held liable along with the users. The new regulations clearly define the conditions for exemption: it’s not enough to simply state that you are not responsible; you must proactively disclose the true nature of your AI module. If someone uses your code for infringement and you were not involved or benefited from it, you can be exempted from liability. This encourages open-source developers to feel more secure and pursue their work without fear of being dragged down by downstream issues.
2. The Current State of Open-Source AI Disclaimers
The disclaimer statements in major open-source AI products are often meaningless. Lawyers analyzed the agreements of several popular models like DeepSeek, GLM, and Kimi and found that they typically consist of just a few vague sentences, such as “The developer makes no guarantees, and the user bears all consequences.” These statements do not address even the basic characteristics of the AI. Even the more comprehensive agreement from Alibaba’s Tongyi Qianwen open-source version uses the standard Apache 2.0 license, which essentially amounts to “I make no guarantees.” This is similar to selling a tool with a disclaimer that says “I won’t compensate for damage,” but not mentioning that the tool cannot be used for dangerous tasks.
3. Common Mistakes in Disclaimers
The agreements of open-source AI vendors all violate the new regulations:
- Passive Disclaimers: Simply stating “I am not responsible” does not meet the new requirements. You must actively list the risks associated with your AI.
- Lack of Clarity: No vendor clearly defines the capabilities and limitations of their AI models. If someone uses an open-source model for purposes beyond its intended use (e.g., autonomous driving or medical advice), the developer may still be held liable.
- Failure to Set Clear Boundaries: Vendors must specify what the model can and cannot do, such as prohibiting its use in military or sensitive contexts. Failing to do so could lead to liability.
4. Why Vendors Hesitate to Update Their Agreements
There are practical reasons for vendors’ reluctance to update their agreements:
- Competitive Pressure: The competition for developers is fierce, and longer, more detailed agreements may deter users. Vendors prefer shorter, more ambiguous agreements to attract users.
- Industry Habit: The software industry has long used traditional licenses like MIT and Apache 2.0, and no issues have arisen. Technological teams are accustomed to using these templates, unaware that they are inadequate for AI.
- Path Dependency: The entire industry has been using these licenses for decades, and no one has realized their limitations in the AI context.
5. The Future of Open-Source AI
With the new regulations, the open-source AI industry is about to experience a wave of compliance updates. Vendors will need to revise their agreements to meet the new requirements. This will benefit both users and the industry as it will reduce the risk of illegal activities and promote more stable development. Users will have better protection, and the open-source ecosystem will be less vulnerable to negative impacts from individual violations.