Apar GuptaUntil a few years ago, the Ministry of Electronics and Information Technology (MEITY/MeitY) was called the Department of Electronics and IT, or DEITY. The abbreviation often led to its overzealous and early attempts to censor the Internet, or regulate technology, open to ridicule. Underlying it is the comic act of government departments that demand divine obedience but do their work with clumsy incompetence. This very avatar was adopted on March 1, 2024, when MEITY issued an advisory to several large platforms for the regulation of generative Artificial Intelligence (AI). It was immediately criticised by both a cross-section of experts in regulation and AI and even the more reticent start-up founders.
An ambiguous stance on legal status
The crux of the issue lies in understanding the legal basis of the government’s text. The term “advisory” lacks definition under the principal legislation empowering MEITY, and the Information Technology Act, 2000 (IT Act). Unlike regulatory bodies such as the Securities and Exchange Board of India, MEITY has no residual powers. But still, it has regularly issued advisories at least since March 2020, with four published on its website until June 2022 and none thereafter. These advisories, aimed at “all social media platforms”, demand vague censorship without citing any legal authority. Contrary to optimistic interpretations, these advisories are more than mere suggestions. One is even titled “notice”, indicating MEITY’s ambiguous stance on the legal status of these advisories. They imply compliance without stating clear penalties. In a weak rule of a country like ours it is common sense for citizens and corporations to perform a compliance charade after an unreasonable demand rather than asking for the official’s rulebook. Hence, these vague advisories are an offer you cannot refuse.
The recent advisory on AI, dated March 1, 2024, marks an escalation following earlier advisories on generative AI (November 7, 2023, and December 23, 2023). Each one of them, triggered by mercurial policymaking, is made at a frantic pace to serve a media cycle rather than be an exercise in sober assessment. In November, the Rashmika Mandanna deepfake video went viral. In December the Prime Minister issued a public warning to “…be very careful before believing the authenticity of a video or an image”. And in February, a Twitter/X user posted a screenshot of Google’s AI response to the question, “Is Modi a fascist?”. MEITY shortly afterwards issued an advisory on each of these occasions. The other common link between all three advisories is the opportunistic transparency adopted by MEITY. Here, their full text has been withheld, with only a press release for the two issued last year. In addition to the press release, there are extensive social media posts and press interviews by the Minister of State, Rajeev Chandrasekhar. These reports show that the November advisory cited social media platforms’ censorship obligations under the IT Rules, 2021. The December advisory reminded platforms to educate users about Indian law under the IT Rules, 2021. In many ways, these were meaningless other than being official coercion for social media companies to perform more proactive censorship.
The March 1, 2024 advisory is a divergence, as Meghna Bal from Esya Center notes, by introducing an illegal AI governance model that requires licensing of AI models. On this, only Minister Chandrasekhar’s press interviews and Twitter/X feed have been the official sources of information. Further, only some Twitter users have published the advisory, indicating selective distribution to media and private firms.
‘Undefined’ terms
This advisory vaguely directs compliance for AI models, including bias prevention and a licensing regime for “under testing” or “unreliable” AI. These terms escape definition under the IT Act, the IT Rules or even the advisory itself. Glaring errors and phrases such as “Indian internet” increase uncertainty.
When many of these issues were highlighted, the Minister’s response was to provide direct interpretation on Twitter/X. On March 4 at 11:43 a.m. he claimed that the advisory “needs to be understood” and abruptly exempted “startups”. Uncertainty was only compounded as none of the terms in his tweet was defined.
Facing growing online criticism and ridicule, the Minister issued his second clarification on Twitter/X at 2:32 p.m., addressing “noise and confusion being created, many by people who shd know better”. This sentence ended with an emoji of a man shrugging his shoulders. But there were signs of a climbdown: “Advisory was simply that — advise”. The Minister avoided any comment on the licensing requirement that goes far beyond the dictionary meaning of “advice”.
Deliberation is old-fashioned now
Dismissing this as just another example of bureaucratic red tape only understates the problem with MEITY’s “advisory regulation”. Technology regulation in India is advancing under questionable legality. The IT Rules, 2021 that were initially intended to regulate user-generated content, have expanded into the “everything law”, as highlighted in the article, “India’s juggernaut of censorship” (The Hindu, January 26, 2023). Its overreach encompasses digital news, video streaming, and online gaming, prompting many High Courts to question its constitutionality.
The practice of issuing advisories without amendments to the IT Rules, but merely making reference to them, represents a further decline in administrative standards. Worse, advisories are subsequently modified on social media posts without any official documentation. This demonstrates how technology policy is increasingly driven by an influencer culture that masquerades short-term ministerial visibility as public interest. Policy decisions are swayed by press coverage and social media metrics rather than the ‘dull and deliberative’ processes of working groups and stakeholder consultations. All of this is rarely put to question and has increased administrative arrogance as a response to government buffoonery.
This was not always the case. As Amber Sinha wrote in Tech Policy Press in 2014, the government quickly retracted a Draft Encryption Policy after online backlash. The times have changed as the recent AI advisory maintains its validity with a muscular if not bungling ministerial defence. Even expert and technical commentary is self censored. Gone are the vibrant and combative challenges that used humour and directed sharp criticism at the highest officers of the Union Government. Developments now signal an environmental change in which there is a greater risk of consequences in questioning policies of digital authoritarianism. The options for us are to pray, be polite and hope for good sense to prevail for there is little else to do. For today, despite an illegal and embarrassing AI advisory, MEITY seems more like a deity.
This article was originally published in The Hindu on March 15, 2024.