Attention Now AI based ‘fraud’ will not work on social media, government issued strict order regarding AI content

By Samira Vishwas On Feb 11, 2026

New Guidelines For AI Content: The Central Government has taken a very strict and historic step regarding the use of social media world and Artificial Intelligence. The Ministry of Electronics and Information Technology on Tuesday issued an order announcing major amendments in the IT Rules (IT Rules 2021), under which now all major platforms like Twitter, YouTube, Snapchat and Facebook will have to mandatorily label every AI content shared on them.

This decision of the government has been taken with the aim of curbing serious problems like misinformation and election rigging spread through deepfake videos and pictures.

These major changes took place in the new rules

According to the new rules, now if any AI-generated content is posted on any social media platform, it will have to be clearly written on it that this content is not original but generated by AI. Additionally, the government has adopted a ‘zero tolerance’ policy against deepfakes. Adopting the policy, it has been ordered that if any objectionable deepfake content is reported, it will be mandatory to remove it within just 3 hours. These revised rules issued by the Ministry will become effective in the entire country from 20 February 2026.

It is noteworthy that the draft of this important change was made public by the government on October 22, 2025, which is now being finalized and implemented so that the digital space can be made secure and transparent.

Accountability of social media platforms

Now under the new IT Rule 3(3) issued by the Central Government social media platforms Accountability has been fixed. According to the new provisions, any platform which provides ‘synthetically generated information’ That is, they will provide the facility to create or share AI based content, they will be given a clear and prominent ‘label’ It will be mandatory to install. The government has clarified that this label will not be just for show, but will be a permanent and unique ‘metadata’ or ‘identifier’ Must be embedded, which no user or platform will be able to change, hide or delete. This technology will help in tracing AI content end to end.

Strict standards set regarding label size also

The rules also set strict standards regarding the size and timing of the label. In case of visual or video content, the AI label must cover at least 10 percent of the screen so that the viewer can easily see it. At the same time, in case of audio content, within the first 10 percent time of the start of the clip, it should be heard that this voice is AI generated.

Along with this, the platforms will now have to develop such advanced technical mechanisms that can identify whether any file is genuine or created by AI even before it is uploaded. This step has been taken to bring transparency in the digital space and prevent fraud through deepfakes.

These 3 changes are also included in the new IT rules

It’s no longer possible to remove or hide labels: Social media companies can no longer allow the removal or hiding of an AI label or its metadata (identifying information). Once the label is in place, it needs to stay that way.
Control on dirty and misleading content: The government has directed companies to use automated tools (software) that can prevent illegal, obscene or fraudulent content created through AI.
Mandatory to give warning every 3 months: Companies will have to give warning to their users at least once every 3 months. They have to be told that if they misuse AI or break the rules, they may have to face punishment or fine.
What will be the impact on users and industry?

Also read: One photo, magic of AI and your cartoon is made, why ChatGPT caricature trend is making waves

Users will now be able to identify fake content easily, misinformation will be reduced. But creators will have to take extra steps, like adding labels. The challenge for the industry will be that they will have to make tech investments for metadata and verification, which can make operations a bit expensive.