‘Most posts flagged as fake news not removed’, shows study
The report also added that the maximum chunk of hate speech was Islamophobic (37%), followed by false news (16%) and posts targeting gender/sexuality (13%) and caste (13%).Updated: Jun 13, 2019 14:05 IST
Hindustan Times, New Delhi
Ninety three percent of all posts reported as hate speech on Facebook in India are not removed and half of those posts initially removed are later restored, a new research report has found, signaling possible loopholes in content moderation.
The report, ‘Facebook India: Towards The Tipping Point of Violence Caste and Religious Hate Speech’, published by US-based Equality Labs, is based on four months of monitoring Facebook posts in India by a team of 20 researchers in four languages — English, Hindi, Tamil and Kannada.
The report also added that the maximum chunk of hate speech was Islamophobic (37%), followed by false news (16%) and posts targeting gender/sexuality (13%) and caste (13%).
The report based its analysis on what Facebook described as Tier 1 hate speech – violent, dehumanising and mocking speech on race, ethnicity, caste, gender, national origin, sex, religious identity and disability, among others.
Such content is liable for immediate removal from the platform.
“The report is a beginning of a necessary conversation to allow more Indians more insight into how Facebook works, how so much hate speech has become normalised, and the categories of hate speech that are now common place,” said Thenmozhi Soundararajan, executive director of Equality Labs.
The report provided sub-categorisation of hate speeches. For example, it found 1/4th of gender/sexuality hate speech was anti LGBT, and half of these called for rape and other forms of sexual violence.
Forty percent of caste-based hate speech attacked caste-based reservations while the rest comprised of caste slurs, anti-Ambedkar messages, among others.
India is Facebook’s largest market with almost 300 million users, and the company has come under fire in recent years for hate and divisive speech on its platform.
In March, in the run-up to the general election, a parliamentary panel on information technology asked the social media giant to ensure that its platform and those of its photo-sharing site Instagram and messaging app WhatsApp were not misused to create divisions or incite violence.
But the report suggests that content moderation needs to be more robust, by developing India-specific community standards, making hate speech and other guidelines available in all Indian languages and augmenting cultural competency of Facebook staff on matters such as caste.
“It is absolutely negligent that Facebook does not have a specific means for reporting casteist hate speech,” Thenmozhi Soundararajan, executive director of Equality Labs, added.
Facebook said it uses technology and reports to identify and remove hate speech and fake news, and that it has introduced caste as a protected characteristics, alongside race, gender etc. “In India, we have long-standing investments in people, technology and partnerships to help us better detect bad content and enforce our policies across multiple languages,” said a Facebook spokesperson in an emailed response.
The number of people working on safety and security issues is at 30,000 worldwide, the statement said, adding more than 15000 of whom are dedicated content reviewers. “We made significant investments during elections to detect and remove fake news from our platform, including increasing our local language capabilities for content reviewers, policies to address the spread of misinformation leading to offline harm and partnerships with third-party fact-checkers,” added the spokesperson.
First Published: Jun 13, 2019 07:09 IST