DeepSeek announced new regulations in China that require all AI-generated content on their platform to be clearly labelled as non-human. So, what should you know about it?
A new measure to prevent misinformation
For your information, the regulations were introduced to increase transparency of AI content and prevent its misuse. All AI content must be clearly marked, in addition to hidden markers in technical data.
Moreover, examples of explicit markers include “AI-generated” text, audio announcements or special graphics. While hidden markers contain information such as the type of content, the company that produced it and a unique ID number. Users are strictly prohibited from changing, deleting or forging these labels.
In fact, any use of tools to cheat the labelling system is also prohibited. Those who violate the rules may face legal action. To support this move, DeepSeek has also published detailed technical guidance explaining how its AI models are trained, the types of data used, and the process of generating content.
What's next?
Additionally, the new rules are in line with Chinese government guidelines that state that all AI content must have a traceable and non-tamperable label. AI service providers are also responsible for ensuring compliance with the rules.
DeepSeek also launched the V3.1 model with 128K context tokens and 685 billion parameters, combining reasoning and generalisation tasks in one model.
Did this news catch your attention? Stay tuned for more news like this at TechNave!
COMMENTS