Ankush Sachdeva, CEO and Co-founder, ShareChat, Moj and Quick TV
The new rules shrinking the timeline to take down flagged content is a bit stringent, and we are engaging with the government on it, Ankush Sachdeva, CEO and Co-founder, ShareChat, Moj and Quick TV, said. “ We are also figuring it out internally. I think it can be better [lesser] than the current timeline [36 hours] but not three, and hopefully we’ll converge on something,” Sachdeva added.
Speaking to businessline on the sidelines of the AI Impact Summit last week, he noted that when it comes to content moderation, the platform has a system that involves both AI and humans. “AI does most of the heavy lifting [in reviewing content] but you want to be absolutely sure before we take down any content. So, AI can probably give you a risk factor, and social media companies don’t want to entirely rely on it,” he said.
With the Ministry of Electronics and IT (MeitY) recently reducing the timeline for removing content upon receiving a government or court order from 36 hours to 3 hours, most social media companies have flagged it as being impractical.
“You have to make sure there are guardrails put in [for AI use] so that any undesirable content doesn’t come in. You cannot use AI to put deep fakes about undesirable content on the platform,” Sachdeva said.
With regard to the impact of AI on social networks and content industry, Sachdeva said that content and coding are probably the two early frontiers where AI is changing drastically. With Gen AI tools, the barriers to creating content gets even lower, so a larger number of people will actually become creators, he noted . “But I think the content that goes viral are the ones which break the pattern; those still have some unique factor working in their favour and mostly involve humans,” he said.
Elaborating on the social network’s own AI endeavours, he traces two time periods in ShareChat’s journey, where they saw AI-driven recommender systems give them a lift in time spent and ad revenue. First when they deployed a deep learning model in production in 2022, which gave them about 40 per cent time spent lift, and the second in June last year, when they saw a 50 per cent lift.
“We have also entered into microdrama as a category and are remodeling all our recommender systems, essentially to better rank microdrama content as well,” he said.
Published on February 23, 2026
