Algorithmic Indecency

Silicon Valley likes to dress its creations in the language of freedom. Artificial intelligence, its apostles insist, is merely a mirror that reflects society’s appetites. India’s brusque 72-hour ultimatum to X over the misuse of its chatbot Grok now punctures that alibi. When a machine repeatedly enables the sexual humiliation of women, such mirrors no longer suffice. It is time to take responsibility.

The Ministry of Electronics and Information Technology’s notice to X is an overdue indictment. Grok, Elon Musk’s vaunted AI companion, has been used to generate and circulate obscene, sexualised and deepfake images of women. Recently, some of India’s brightest women cricketers fell prey to Grok’s indecency.

The danger lies not merely in the volume of abuse, though that is alarming enough. AI-generated images now flood social media with the ease of spam. Anyone online can be targeted and women in the public eye are especially hunted. The result is a grim democratisation of misogyny.

The Ministry has accused X of failing its statutory due-diligence obligations under India’s IT Act and rules, and of neglecting mandatory reporting under newer criminal statutes. It has demanded an action-taken report, a review of Grok’s technical and governance frameworks, and a strict enforcement of user policies and immediate takedowns. It has threatened the ultimate sanction - the loss of “safe harbour” - the legal shield that protects platforms from liability for user content.

Grok has already made headlines in India for abusive language and reckless forays into politics and history, dispensing opinions on sensitive figures and controversies with adolescent bravado. That a chatbot can declare one politician “more honest” than another, boast that it is “not afraid of anyone” and then be weaponised to strip women of dignity points to a deeper flaw.

The fact that women are compelled to negotiate their bodily autonomy with software is a dystopian footnote to the AI age. Consent, once a social norm enforced by law, is being reduced to a disclaimer ignored by bad actors and insufficiently policed by platforms.

Platforms that deploy generative tools at scale assume a duty of care proportionate to the power they unleash. India’s response has been notable for its sternness in signalling that AI-enabled obscenity will be treated as an out and out crime.

Laws on indecent representation, child protection and criminal conduct already exist. The question is whether technology firms will align their incentives with those laws or continue to test how far they can go before being stopped.

Grok’s promise was to be a witty, unfiltered companion. But it seems en route to perilously becoming a veritable factory for harm by laundering harassment through code. The choice before X is to either build serious safeguards and accept accountability or lose the privilege of operating behind legal shields. The age of AI bravado is ending. What comes next must be responsibility. By turning harassment into a prompt-response service, X has not liberated expression but mechanised cruelty.