Meta said on Tuesday that it has changed the systems that decide what Instagram recommends in India so that more sexually suggestive and violent content is kept out of recommendations. It said people now report recommended content about half as often as they did a year earlier. Meta did not say how much content was kept out or how many reports it received a year ago.
How it says it works
Meta said its AI models now assess content directly instead of waiting for a user to report or hide a post, and that it tuned the systems for India. It said its systems worldwide check content against its recommendation guidelines, that content breaching its community standards is removed, and that content unlawful in a country is restricted there. It listed user controls including a sensitive content control, a political content control, hidden words, block, and teen accounts, which place users aged 13 to 17 in a 13-plus content setting by default and were introduced in India last year. It pointed to its global enforcement report and its monthly India report under the IT Rules, 2021.
The government context
A week earlier it was reported that Meta made similar claims at a 21 August meeting with the home ministry's Indian Cyber Crime Coordination Centre, telling it that objectionable AI-generated content was no longer being recommended and that safeguards against adult nudity had been strengthened, again without figures. The centre recorded that Meta's enforcement so far remained insufficient and asked for a briefing on the principles behind its algorithms. The minutes were filed in the Supreme Court on 7 October in a case on digital arrest scams. Meta's Tuesday statement does not mention that meeting, and its claims are its own and have not been independently verified.









