Premium
This is an archive article published on November 24, 2022

Meta transparency report: Decrease in bullying, harassing content on Facebook and Instagram

According to Meta's Integrity and Transparency Reports Q3 2022, there is a surge in user data requests from the Govt of India.

Meta has improved AI capabilities to detect bullying and hate related cotent on Facebook and Instagram (Photo credit: Reuters)Meta has improved AI capabilities to detect bullying and hate related cotent on Facebook and Instagram (Photo credit: Reuters)

Meta has published its quarterly integrity and transparency report for third quarter of 2022. As per the report, there is a sharp decline in hate speech-related content from 13.5 million to 10.6 million on Facebook in the quarter. The social media conglomerate claims that it has improved the accuracy of its artificial intelligence technology using data from previous user appeals and is now able to distinguish between actual hate speech and humorous terms used between friends.

In a blog post, Guy Rosen, VP of Integrity at Meta wrote that the company can now “identify posts that could have been removed by mistake without appropriate cultural context,” as they can detect “better detect words that may be considered offensive or inappropriate in one context but not another.”

The report also suggests that Facebook has improved the “proactive detection rate for hate speech” from 95.6 per cent to 90.2 per cent. Proactive detection means metric “the percentage of all content or accounts acted on that Meta found and flagged” before users reported these to the platform. This metric is an “indicator of how effectively” Meta’s platforms detect violations.

Story continues below this ad

It also claims to have reduced the proactive rate of bullying and harassment-related content from 76.7 per cent to 67.8 per cent on Facebook and 87.4 per cent to 84.3 per cent on Instagram.

But it should be noted that while there is a decrease in the proactive rate, the amount of harmful content remains the same on both platforms. Meta does indicate that with the improvement in AI technology, this number is expected to come down in the coming quarters. Meta has taken action against 2.2 million pieces of terrorism-related content this quarter. With the help of improved proactive detection technology, Facebook has also identified and removed over 2.5 million pieces of drug content.

Meta transparency report: Requests from India

According to the report, the Indian government sent 55,497 requests to Meta, of which 51,602 are legal requests and 3,895 were emergency disclosure requests. Meta provided some data to 66.59 per cent of these requests in 2022 and that there continues to be an increase in user data requests from India. In comparison, the total number of requests from India, in the second half of 2021 stood at 50,382.

Latest Comment
Post Comment
Read Comments
Advertisement
Loading Taboola...
Advertisement