Privacy experts wary of covid directive

The government, in a letter to social media companies last week, said the World Health Organization (WHO) had classified the variant as B.1.617. The term ‘Indian variant’ was without basis, it said.

However, tech policy analyst Prasanto K. Roy pointed out that the Indian government itself has referred to earlier variants of the virus by the name of the countries they were found in, such as the UK strain and the Brazil strain.

A keyword-based ban such as this could affect posts that have no offensive content, Roy said. “A keyword-based ban is over the top and disproportionate and you could take down a whole lot of other posts as collateral damage. The last time they did this, they ended up taking down tweets of supporters of the ruling party who were flagging other tweets with the allegedly offensive hashtag,” he said. “A sweeping ban like this is rare, except for perhaps in China and Russia. China, in fact gives keyword based, real-time takedown directives even to messaging platforms,” he said.

“This is one of those things where Twitter and Facebook want to do it algorithmically, but they can’t, because the algorithm doesn’t get context. If you give it a term ‘Indian variant’, it will remove every mention of that whether it relates to covid or not,” said Alok P. Kumar, founder, Vidhi Centre for Legal Policy.

“The government of India has no legal authority to tell people what terms they should use to describe a virus. They have no legal authority to tell Facebook or Twitter to police speech, which is not prohibited under any legislation in India,” Kumar said.

A directive like this would require platforms such as Facebook and Twitter to not only look at posts within India, but also posts originating outside the country, according to experts. They will have to block all these posts from showing up in the country.

This is different from banning a hashtag, where social media companies can simply ban posts that use a particular hashtag on a wholesale basis. In the case of keywords, they have to figure out the context, which algorithms cannot do.

Companies such as Facebook and Twitter could try to teach their artificial intelligence and machine learning algorithms to sort out such posts contextually, but such efforts haven’t been very successful. In fact, it could lead to inequality, where posts by accounts with high follower counts and those of dignitaries are referred to human reviewers, while regular users’ posts are taken down without any recourse. Neither Facebook and Twitter commented on this.

The government is well within its rights to ask fake news to be taken down, said Supreme Court lawyer N.S. Nappinai, the founder of Cyber Saathi, a cyber safety initiative that disseminates information on cyber laws. However, Nappinai cautions against the use of blanket bans, as the first reaction to any problem. Advisories flagging fake news themselves would have helped, she said.

Nappinai advised that the government ought to look for alternatives or proportionate action. “When it comes to issues of free speech in general, it would help if the first reaction is to choose an effective remedy that does not restrict fundamental rights,” she said.

The government of Singapore had, on 20 May, directed social media firms to carry a “correction notice” to end users in Singapore, who post content related to a Singapore variant of the virus. This came in the wake of Delhi chief minister Arvind Kejriwal announcing the existence of such a variant on social media.