Facial recognition has moved in India from pilot projects to routine policing without a statute that says when, against whom, and for what purpose a face may be scanned. This paper examines what that gap means when the face belongs to a person attending a political protest. Attendance at a demonstration is a public act, but the record of who attended, kept by the State and matched against police databases, becomes a durable data set from which political affiliation can be inferred, whether or not the police intend to infer it. The paper builds its argument around the events at Jantar Mantar in July 2026, when the Delhi Police deployed a facial recognition system during the student agitation led by the Cockroach Janta Party over the NEET-UG paper leak, reported that it had matched 2,873 people against its criminal records, and defended its conduct before the Delhi High Court and the Supreme Court. It traces the litigation that followed, including the surveillance petition before the Delhi High Court, the petitions against the police action of 20 July before the Supreme Court, and the Supreme Court’s order of 1 September 2026, and it records what each court said and what each court left open. It then compares India with the European Union, the United Kingdom and the European Court of Human Rights, each of which has treated protest-site biometric identification as a distinct problem. The paper contends that the Indian position is not a case of unlawful technology meeting a silent Constitution, but of a settled constitutional standard under Puttaswamy meeting an administrative practice that has never been tested against it. It proposes a narrower and sequenced framework built around a dedicated statute, prior authorisation, a bar on inference of political belief, independent oversight, auditable accuracy standards, and protest-specific rules on retention and notice.