We are witnessing significant advancement in AI technology especially in the areas of text to image and text to video generator with simple prompts ouputing results that are at par (even above-par sometimes) with those same content produced by the humans.

China has expressed concern in AI content produced by digital platforms and has introduced guidelines compelling them to provide information about content generated through AI. The guideliness still in its draft stages has call for opinions from the public till October.

If the law gets passed, it would mandate the netizens taking help of AI to clearly let the public know about it.

According to the draft, AI generated content can have a logo appear in the content which would represent that AI has been used to create it.

This proposal comes at a time when there is a significant rise of AI generated content on the Internet. The extent to which AI can be used by bad actors can be easily understood by deepfakes of noted celebrities.

As per the draft, platforms who will fail to comply by the guidelines can face stringent actions.

For AI generated text content, the proposal calls for adding a logo, or adding a text prompt at the beginning, end and at other appropirate timelines.

For AI generated video content, the draft says that prominent warning signs can be added at the beginning of the video and at appropriate locations around the video.

For the AI generated audio content, prominent voice prompts or audio rhtym prompts need to be added at the beginning, end or at appropriate positions. Similarly, prominent prompt signs should be provided on the interactive scene interface.

Adding Identifiers to the Metadata of AI generated content

The draft also proposes that service providers should also add implicit identifiers such as metadata to the synthetic generated content. They are also encouraged to add a digital watermark if they generate synthetic content. Metdata can be hard to interpret and even tampered with and in many cases require some technical expertise to view it.

Provision of using AI content without Identification

There may be scenarios where an organization may be required to use AI generated content without providing explicit information. In such case, the organizations or end user can do so through a user agreement with the AI service provider. The agreement will have details of the user and the service provider will have to retain the user information log for atleast six months.

The draft is open to public opinions till 14th October 2024 whereas the likely date for the law to come into existence is Jan 2025.

While the laws may look stringent, they could mitigate the adverse effects of AI by bad actors and other countries may as well take help from it in order to create stringent laws for their nation.