Editorial Team
7 min read
Delhi Police says its live facial-recognition deployment at Jantar Mantar is meant to find wanted people and prevent disruption at a large student protest. The public record still does not show the current match threshold, the legal basis and scope of the watchlist, how long footage is kept or the safeguards after an alert.
Delhi Police, the force responsible for policing India’s National Capital Territory, is running live facial recognition on video feeds around Jantar Mantar, where the Cockroach Janta Party (CJP), a student protest group, has held a sit-in and hunger strike over repeated examination-paper leaks since June 20.
Police say the system is not directed at ordinary demonstrators. It is meant to flag wanted people, absconders, history-sheeters and others whose images are already in police records, and to keep criminal or anti-social elements from exploiting a large gathering. That explanation has a concrete immediate context: a report based on police sources says scores of protesters and police personnel were injured in clashes at the site on July 20.
But the use of a live matching system at a protest means the public question is not only whom police say they want to find. It is also what information is taken from everyone who passes the cameras, the database against which it is checked, and the limits on a potential match. The current reporting does not publicly establish those controls.
The accounts describe two vehicles parked outside Kerala House near the protest: a Mobile Command and Control Vehicle, where officers view live CCTV feeds, and Ikshana, a mobile platform whose feeds are processed through facial-recognition software. Ikshana was inducted before Delhi’s 2023 G20 Summit. It has eight fixed cameras providing 360-degree coverage; its stated functions include face detection and recognition, number-plate recognition, traffic management, crowd monitoring and video analytics.
Police sources said four FRS units were at entry and exit points, directly connected to the Delhi Police database. A separate police account described four FRS-equipped vehicles in and around Jantar Mantar, and said nearly 100 facial-recognition cameras are installed across New Delhi district. The published accounts do not explain how those counts relate to the two vehicles or which camera feeds are matched, so they should not be read as a single confirmed equipment total.
In the account of the four vehicles, officials said the system compares faces against a database of more than 200,000 suspected criminals and had identified at least 60 people whose faces matched records and who were suspected of criminal backgrounds. That is a police claim, not a disclosed measure of confirmed identifications: the account does not state how many people were scanned, how many alerts were reviewed or confirmed, or whether the 60 figure covers Jantar Mantar rather than the wider district.
Police say facial recognition has been used for years at major gatherings including Republic Day, Independence Day, Holi, Diwali, Eid and Ramlila. They also say central security agencies have provided intelligence that cross-border terrorist organisations may seek to exploit the student protest. That account of the police case does not identify a site-specific threat assessment or independently verify a risk at Jantar Mantar. It does, however, show that the deployment is being presented as part of a broader public-order practice rather than as a new technology introduced solely for this protest.
Facial-recognition software converts features from an image or video into a template and compares it with stored images. A similarity score that clears a threshold produces a potential match, not proof of identity; the result can vary with the algorithm, lighting, camera angle, image quality, masks and the database searched.
Delhi Police said in a 2022 RTI response that it treated a similarity score of 80% or more as a positive result. The contemporaneous account of the current deployment does not establish whether 80% remains the threshold, which algorithm is being used at the protest, how an officer verifies an alert or the system’s performance there.
The same account says police have not publicly detailed the size or composition of the current database, the software’s accuracy at the site or the procedure after a possible match. An officer also said there were no fixed retention guidelines for protest footage because it might be needed in a future investigation. Those gaps matter independently of the police rationale: a system can be intended to identify wanted people and still require clear limits for people who do not match a record.
Internet Freedom Foundation illustration accompanying its July 24, 2026 representation and RTI applications concerning live facial recognition at Jantar Mantar. Source: Internet Freedom Foundation.
The Internet Freedom Foundation (IFF), a digital-rights organisation that has pursued RTI disclosures about Delhi Police facial recognition since 2020, says replies it obtained after a Central Information Commission appeal showed that the force had no specific rule for the technology in 2022 and had conducted no privacy impact assessment. Those are historical disclosures, not proof that no later rule exists.
IFF’s July 24 representation asks the Commissioner of Police to halt live facial recognition at peaceful assemblies unless there is specific legal authorisation; to delete footage, facial templates and match results for protesters unconnected to a specific cognisable offence; and to publish operating, retention and destruction rules. Its four new RTI applications seek, among other things, the current threshold, accuracy audits, legal basis for the databases, deployment orders and instructions governing CCTV and videography at the protest.
The group’s position is contested. It argues that live matching at a lawful assembly has no adequate legal basis and is disproportionate; police describe the system as a targeted security measure. The legal setting is also more qualified than either slogan: the reported legal analysis notes that India’s Supreme Court recognised privacy as a fundamental right in 2017 and requires restrictions to meet legality, necessity and proportionality tests, while the Digital Personal Data Protection Act, 2023 contains exemptions for some law-enforcement processing and no dedicated police-FRS operating regime.
Former Jawaharlal Nehru University Students’ Union president Aishe Ghosh, now the petitioner challenging the surveillance of the CJP protest, has alleged continuous and indiscriminate monitoring in the Delhi High Court. At a July 20 hearing, Solicitor General Tushar Mehta, representing Delhi Police, said videography was for law and order and public safety rather than snooping. The court asked whether the force had a standard operating procedure or standing order for handling protests in line with earlier Supreme Court directions.
CP PLUS, the surveillance-technology brand of Aditya Infotech Limited, said it designed and developed the Ikshana prototype with Delhi Police and integrated its Health Monitoring System software for live monitoring of the surveillance setup. That gives the company a material role in a police platform now being used at a protest; it does not establish that the company can access protest footage or biometric data.
IFF says no tender, contract or data-handling agreement for Ikshana has been disclosed and has asked whether vendor software sends information to private servers or permits private access. Those are questions in an RTI request, not findings. The available record supports a narrower conclusion: the public has not yet been shown the contractual and data-governance terms of the vendor relationship.
The next test is documentary, not rhetorical. Delhi Police could substantiate its claim of a bounded security tool by releasing the current database categories and legal basis, match threshold, accuracy and audit information, alert-review protocol, retention and deletion schedule, site deployment order, and the Ikshana vendor and data-access terms.
Those records would also let the public assess the trade-off between the force’s stated aim of locating people already sought by police and the cost of scanning people who have come to a public assembly. Until then, the evidence establishes both that facial-recognition is live at Jantar Mantar and that the rules governing its use there remain materially incomplete in public.
Sundar Pichai said Google supports an industry letter on open models, aligning the company with developers and technology firms resisting broad limits on Chinese AI systems while leaving allegations of IP theft and national-security risks unresolved.
Brazilian courts and bar associations have responded to hidden prompts in legal filings with fines, referrals, suspensions and an investigation. The known cases were detected, however, and do not show that an AI system changed a judicial decision; they instead test how courts secure AI-assisted review of untrusted case documents.
Salamanca City Central School District has put its proposed Realbotix humanoid robot and Optio AI assistant pilot on hold while it works on student-data privacy agreements. The pause leaves the district with a more basic task: establish what the digital tutor and the physical robot each collect, do and improve before either reaches students.
Samsung Electronics and Broadcom have signed an MOU covering memory, foundry technologies and advanced packaging for next-generation AI infrastructure. The companies expect more than $200 billion of collaboration over five years, but have not disclosed how that figure translates into purchases, production or capacity commitments.
OpenAI says GPT-5.6 Sol reached a 24.9% ExploitGym pass rate under a two-hour cap, up from GPT-5.5’s 15.1%. The benchmark measures a bounded step from a known vulnerability to an exploit; a separate assessment performed with OpenAI finds continued limits on hardened targets and long operations.
Chey Tae-won, chairman of SK Group, said Anthropic had sought SK Hynix component-supply cooperation for chips it hopes to make itself. But Anthropic has disclosed no chip plan, and the same summit also produced a separate letter of intent between Anthropic and SK Telecom on a large AI data center.
South Korea announced $950 billion in planned five-year semiconductor cooperation involving Samsung, SK Group and global technology companies, alongside prospective AI-data-center projects. The figure aggregates arrangements with different terms and disclosed commitments, leaving delivery, power and the overlap between the SK and NVIDIA figures unresolved.
OpenAI said it applied a fix after a July 25 disruption affected ChatGPT, its Codex coding service and parts of its API. The available reporting points to recovery in about 50 minutes, but the retained incident records do not explain the cause, the mitigation or whether the services failed through a common component.
Anthropic CEO Dario Amodei said the Claude developer has signed supply deals with Samsung Electronics and SK hynix. The available disclosures show a useful infrastructure relationship, but not an HBM purchase, a dedicated allocation or a claim on either supplier’s output.
Moody's Ratings expects six US hyperscalers to spend $785 billion in 2026 and close to $1 trillion in 2027. The forecast signals continued demand for AI capacity, but a separate Moody's analysis shows why capex alone understates the financing question: five of the companies already reported $969 billion in undiscounted future lease payments, much of it not yet on their balance sheets.
Anthropic has released Claude Opus 5 at Opus 4.8's API price and says it approaches Claude Fable 5 on coding and knowledge work. An Epoch AI listing places it at 159 ECI, below Fable 5 and GPT-5.6 Sol, leaving customers to test whether the cheaper model delivers better economics in their own workloads.
Nvidia plans to invest $1 billion in Naver as Brookfield signs a nonbinding term sheet for up to $9 billion toward a South Korean AI-factory project. The companies target a 200-megawatt expansion by 2028, but financing, definitive terms and construction remain ahead.
Nvidia and SK Group signed letters of intent for an AI-infrastructure initiative they value at more than $500 billion, centered on an up-to-2-gigawatt Korean AI factory and a long-term SK hynix memory partnership. The disclosed projects give the plan a technical shape, but not a capital breakdown, time horizon or basis for treating the headline figure as committed spending.
Debian has opened discussion of two General Resolution choices: a ban on direct LLM-assisted contributions that would amend its Social Contract, and a conditional-use policy that puts responsibility on contributors. The official text shows that neither policy has been adopted and that the permissive option has changed from its initial mailing-list draft.
Exposed attacker infrastructure contained logs showing the Hermes AI agent running unattended reconnaissance and privilege-escalation checks against systems associated with Thailand's Ministry of Finance. The artifacts show deliberate, operator-directed targeting, but they do not establish the initial access path, the full extent of compromise or data theft.
Amazon is closing its San Francisco AGI Lab amid undisclosed job cuts, while saying its frontier-model research under Pieter Abbeel and its Nova Act service will continue. The closure follows leadership departures and an AGI reorganization; Amazon has not said how many people are affected or where the lab’s work will go.
Vice Premier Ding Xuexiang’s visits to Huawei and computing facilities connect China’s domestic AI-chip push to a national effort to schedule compute, data and power. The harder test is whether those systems can make domestic hardware efficient and dependable while Chinese developers and cloud rivals reduce their reliance on Huawei.
The Delhi High Court has denied ANI Media's request to immediately stop OpenAI from using its news material to train the models behind ChatGPT. Its prima facie fair-dealing finding gives OpenAI interim relief, but leaves the copyright suit, the evidence and the question of compensation unresolved.
A viral reel recorded at a Delhi trans-rights protest with Ray-Ban Meta glasses was initially left up by Instagram and later removed. The episode shows how a recording indicator and user rules can fall short when people filmed without consent must discover the camera, navigate moderation and contend with copies after the fact.
CXMT reportedly charged more than Samsung for comparable DDR5 server modules and signed large supply agreements with ByteDance and Tencent. The reported premium reflects tight supply and state-shaped domestic demand, while CXMT remains behind leading rivals in high-bandwidth memory and lithography capability.