With the meteoric advancement of artificial intelligence (AI) technology in China, Deep Seek and Manus are now the latest tools under the spotlight. Businesses, employees and contractors are keen to utilize such technologies in the workplace to enhance competitiveness and efficiency. In this newsletter, we answer the burning questions from our clients on utilizing AI in their businesses.
Q0. What is DeepSeek and Manus?
DeepSeek is a chatbot which has an interface similar to ChatGPT. Deepseek has a reasoning model, before delivering the final answer, the model first generates a chain of thought. Users may view the chain of thought to further modify the parameters and obtain a more accurate response.
Manus AI is a general AI agent designed to be an autonomous system that perform and execute tasks. Manus differs from chatbots that primarily generate ideas or assist with specific workflows, it is designed to independently perform tasks from start to finish.
Q1: Can I claim ownership of the content I created using AI technologies and stop others from using my AI-generated contents?
Latest case laws in China have recognized that AI-generated work is copyrightable. In the latest case on copyright of AI-generated pictures, the Suzhou Changshou court that the plaintiff’s modification of the prompt and modification of image details through image processing software reflected his unique selection and arrangement, and the generated floor plan was original and was a work protected by the Copyright Law. Also, under the case law, where the work was created due to the input of development teams in a corporation, the corporation/ employer was found to be the author.
However, as of now, whether or not a specific AI-generated work is copyrightable in China will still need to be determined by the court on a case-by-case basis. AI-assisted work (as opposed to autonomously generated work) will more likely be considered copyrightable. The involvement of human intervention, choice and judgment in relation to the outcome of the AI-generated work is vital. The copyright ownership would also depend on the copyright clause in the user agreement with the AI service provider.
To claim copyright ownership on AI-generated content, it is advisable to keep evidence of inputs, prompts and modifications. You may even consider filing a copyright recordal in China to keep as evidence.
Q2: What are the legal risks my company have to be aware of when utilizing AI technologies
The Chinese courts have found AI-generated content constitute infringement. Problematic AI-generated content has been found to be constituting unauthorized derivative works to the prior copyrightable works. In the Ultraman case, the Guangzhou Internet Court found the defendant, a text-to-image AIGC provider, liable for infringing the copyrights of the famous Ultraman IP. Problematic AI-generated voices were also found to be violating personal rights related to voices.
Privacy considerations in prompt input and learning data training are also another risk area. You should also understand how the AI service handles data and privacy. Understand what data is collected, how it is used, and whether your inputs or generated outputs might be stored or used for further training of the AI.
Attention must be paid on the consequence of prompts and learning data which you input, whether they would be incorporated into the AI tool and would be accessible to other users of AI tool that are not from your company. Also, be aware of how the service can terminate your access and what happens to your generated content and the data which you input upon termination. A thorough review of the terms and policies of the AI tool is vital before adopting the tool to ensure compliance with internal policies and relevant data protection laws.
Policy has to be in place on restricting confidential data to be used in AI tools. Where customer data would be further processed by AI tools for further business enhancement, you would need to review the data collection policy to ensure proper consent is obtained.
Q3: What legal remedies are in place for businesses whose data is used by generative AI providers?
In the Ultraman case, in view of helping AI users safeguard their IP rights, the court emphasized the following essential responsibilities of AIGC providers:
- Promptly halting infringement activities and removing relevant materials from the training database to prevent ongoing violations;
- Implementing keyword filtering measures;
- Providing clear identification notices to prevent infringement; and
- Enhancing complaint and reporting mechanisms.
Businesses should be aware of the complaint system and policies of generative AIGC providers, and should promptly report to AIGC providers should infringement be found. Businesses should develop a watch on popular AI tools to check if its proprietary information and works can be generated in the system.
Where co-operation with third parties involve transmission of data, clauses should be in place in restricting the use of such data, to prevent the data to be sold or inputted into AI tools which may lead to the development of AI-infringing content.
Q4: What policies should my company put in place?
In relation to utilizing existing AI tools, you need to put in place a restriction of usage, defining what tasks and information can utilize AI tools. You should review the terms of conditions of the AI tools used by employees and contractors, see if the tools have any IP and privacy safeguards on the legality of data collected, trained and used. You may then provide a list of AI tools that may be used and review the list from time to time. If other AI tools are to be used, you should evaluate compliance and risk.
It is important to set out IP guidelines in use of AI tools, including forbidding use of infringing materials or prompts. Where the AI tool allows for input of training data, care should be taken into the legality of the training data. You should also use plagiarism detection tools to ensure the AI-generated content does not closely resemble any existing works. You need to evaluate the generated content for potential biases, inaccuracies, or harmful information, and ensure that it aligns with ethical standards and your business’ values. Any complaint by third parties on infringement of AI-generated content should be dealt with promptly and infringing contents be removed.
We hope the above answered your questions on these exciting tools.
