amnesty

New Delhi, India – An investigation by human rights group Amnesty International has confirmed that Indian government forces used lethal weapons, including pellet-firing shotguns, grenades, batons and electric shock devices against demonstrators in Gen Z protests that rocked the national capital last month.

On July 20, thousands of protesters had all but shut down New Delhi, blocking off Parliament Street, demanding the resignation of then-Education Minister Dharmendra Pradhan following a string of examination scandals involving the leaking of national exam papers and other irregularities.

More than 100 students were injured in the crackdown by police and paramilitary forces that day. Indian newspapers obtained hospital records that showed that doctors had treated pellet wounds in several cases.

India’s opposition, led by Congress party leader Rahul Gandhi, has since called on Home Minister Amit Shah, seen as Prime Minister Narendra Modi’s confidant, to answer questions about the alleged police brutality and has joined protesters in calling for his resignation.

But despite mounting evidence and growing political criticism, the Modi government repeatedly insisted that security forces had been measured in their response to the protests and denied using excessive force.

Now, an Amnesty investigation has found the use of tactics that Indian law enforcement rulebooks themselves warn against using in all but exceptional circumstances.

What happened on July 20 in New Delhi?

In July, the Cockroach Janta Party (CJP), a satirical, youth-led movement which gathered momentum in response to the national examinations scandal, called on its supporters to march on the Indian Parliament after a monthlong sit-in failed to oust the education minister.

Two days before the march, the police forcibly removed renowned activist Sonam Wangchuk, who had been taking part in a hunger strike. That move galvanised public anger, bringing out more protesters in New Delhi.

On the morning of the march day, the police used tear gas as protesters swelled, then attacked them with electric shock batons, as well as shooting at them with pellet guns, Amnesty International’s Evidence Lab has verified from video and photographic evidence alongside witness testimony.

Amnesty also found that government forces had put up barricades and enforced a communications blackout.

“The weapons were deployed in a manner that breached international law and standards and domestic policing guidelines,” Amnesty said in a statement on Monday.

The rights group’s report also noted that the Delhi Police had publicly denied any disproportionate use of force, specifically pellet guns. The police also repeated this before India’s top court during a hearing about the alleged use of excessive force.

Amnesty says the evidence it has gathered “directly contradicts [the police’s] account”.

Four days after the July 20 march, as protests spread across the country, police in Bihar state’s Siwan town opened fire on protesters, injuring at least three youths.

Did government forces fire pellets at protesters?

Following the July 20 protest, several videos and photos emerged of protesters with injuries consistent with pellet wounds, which were reflected in their medical records as well.

The Evidence Lab verified two videos showing a paramilitary Rapid Action Force officer firing a shotgun into the crowd at the intersection of Connaught Place and Parliament Street in New Delhi, where protesters had gathered.

At least two protesters had wounds consistent with those caused by birdshot, a hunting round that fires a large number of small metal pellets, the group noted.

“The doctors confirmed that they were pellet injuries. Thankfully, my bones have been saved, but my body has been scarred badly,” one protester told Amnesty.

Amnesty said birdshot ammunition has not been developed for law enforcement purposes due to the widespread dispersion of the metal pellets, which “is inherently inaccurate and presents a high risk of serious injuries to the targeted person and even bystanders”.

Amnesty also said the police did not issue a public warning before firing pellet guns, as recommended by the Indian government’s Bureau for Police Research and Development.

Police and security forces have used metal pellets for more than a decade in Indian-administered Kashmir, leading to the first mass blinding during a 2016 civilian uprising.

Posting on X, Delhi Police called the reports of pellet injuries during the latest protests “fake news”, “completely false and misleading” and warned that “appropriate legal action” would be taken against anyone “spreading rumours”.

A subsequent internal inquiry by India’s paramilitary forces, however, reportedly found that personnel had fired at least seven rounds of ammunition containing metal pellets.

“Instead of facilitating the right to protest, Indian authorities suppressed it first with a communication blackout, barricades and transport disruption. They then attempted to crush it with unnecessary or excessive use of force against peaceful protesters, including children,” said Aakar Patel, chair of the board, Amnesty International India.

“This heavy-handed response was state-sanctioned violence masked as crowd control. The continued impunity after a month is a testament to that,” Patel said.

Did police use tear gas grenades and batons?

Government forces did use tear gas grenades to disperse the crowd at several places in New Delhi, as seen by Al Jazeera and verified by Amnesty, not just on July 20 to prevent CHP supporters’ march to Parliament.

“Tear gas grenades releasing an excessive amount of explosive energy are extremely dangerous and should be prohibited for use in public assemblies,” Amnesty said, referring to videos of tear gas shells exploding near students.

The group also noted that witnesses observed that “tear gas grenades were launched directly at protesters instead of at an angled trajectory above the head of protesters, at odds with UN Guidance on the use of Less-Lethal Weapons in Law Enforcement”.

Amnesty has also verified claims that the police used lathis, the Indian version of batons, to beat protesters.

The UN special rapporteur on torture considers lathis to be inherently cruel, inhuman or degrading as “they deliver a greater level [of] kinetic force than conventional batons, causing excessive pain and increasing the risk of serious injury” and has called for them to be prohibited, a position that Amnesty supports.

The group also found men in civilian clothing beating protesters “while uniformed officers, who escorted them toward the crowd, stand beside them and make no attempt to intervene”, as visible in videos seen by Al Jazeera. Several officials were not wearing their name tags or badge numbers, as required.

What has been the reaction to the use of excessive force?

A day after the crackdown in New Delhi, CJP founder Abhijeet Dipke told the crowd, “Every police person who attacked you would be identified and dragged to the courts”.

The group also launched a platform for protesters to upload videos and photos from the crackdown.

The opposition, led by the Congress’s Gandhi, pushed the home minister for a discussion on the issue in Parliament, an issue that plagued the monsoon session with Shah missing.

Gandhi argued that either Shah ordered the usage of lethal weapons, in which case “he is guilty”, or he was unaware of the situation, which makes him “incompetent” for the job.

Sahil Lochab, a 19-year-old who lost vision in one eye due to pellet injuries, said the police had been refusing to register his complaint. In response, Gandhi staged a seven-hour sit-in outside the office of the Delhi Police chief, insisting that a case be registered over the use of pellet guns on protesters, alongside Lochab.

After Gandhi’s pressure, the police relented. A student at the University of Delhi, Lochab received multiple pellet injuries on his chest, back, arms and face. Several pellets could not be taken out, he said in his complaint.

“My medical documents show that there are over 200 pellets in my body. Doctors have told me that the possibility of my eyesight coming back is less than 1 percent,” Lochab wrote in his complaint.

The Question Islamabad Has Not Answered

On the night of 16 March 2026, Pakistani warplanes struck a walled compound in Kabul’s ninth police district. Afghan officials say more than four hundred people died. The United Nations mission in Afghanistan has independently confirmed at least two hundred and sixty nine deaths and some one hundred and twenty injuries. Islamabad’s account was simple: the site held ammunition and military infrastructure, and any civilian harm followed from secondary explosions rather than the strike itself. Four months later, Amnesty International published an investigation reaching the opposite conclusion, built not on Taliban testimony but on satellite imagery spanning three years, verified video, and independent geolocation of the wreckage. On 9 July, Amnesty asked Pakistan’s foreign ministry for the evidence behind its military target claim. As of publication, no response has arrived. That silence, more than any single photograph, is now doing the analytical work in this dispute.

A Conflict Escalating Since October

The Omid strike did not occur in isolation. Cross border hostilities between Pakistan and the Taliban administration have intensified since October 2025, and the UN mission in Afghanistan has documented more than seven hundred and fifty civilian casualties in the first three months of 2026 alone, spread across a series of Pakistani air operations against targets it describes as militant infrastructure. The Omid center, a drug rehabilitation facility expanded after 2021 to house roughly two thousand patients, was neither obscure nor newly built. Its treatment programs featured in a 2023 Al Jazeera documentary, and signage identifying the site as a medical facility was visible above its main gate. This is the backdrop against which the competing claims about the March strike have to be weighed. It is a theatre where verification has become as contested as the fighting itself, and where each side’s evidentiary standard has become a proxy for its wider legitimacy claim.

How Open Source Evidence Reshaped the Dispute

The strongest version of Pakistan’s position would rest on operational intelligence unavailable to outside observers: a genuine military use of the site that satellite imagery and testimony could not capture, perhaps weapons storage concealed within a facility that also treated patients. That is a coherent claim in the abstract, and cross border conflicts involving non state or quasi state administrations often do see dual use of civilian infrastructure. The problem for Islamabad is not that this claim is inherently implausible. It is that four months of opportunity to substantiate it, including a direct written request from an investigating body, has produced silence rather than documentation.

Set against that silence is a body of evidence that does not depend on Taliban cooperation. Amnesty’s investigators analysed satellite imagery of the compound from April 2023 through 15 March 2026, tracking the site’s physical layout and, in imagery from just three days before the strike, the presence of large numbers of people in the patient and administrative areas. A video recorded in 2025 and verified by the organisation shows the interior housing arrangements for patients. Separately, and this is the detail that most changes the analytical picture, broadcast journalists reviewing Pakistan’s own released strike footage identified at least four buildings struck with guided munitions, while a conflict researcher affiliated with the open source investigation group Bellingcat independently geolocated the strikes to the rehabilitation center’s buildings using imagery unconnected to either Amnesty or the Taliban administration.

This matters because it breaks the framing in which the dispute is simply Pakistan’s word against an adversary regime’s word. Geolocation from Pakistan’s own footage and independent satellite analysis are not Taliban sources, and they are not filtered through UNAMA’s access constraints in Afghanistan, real as those constraints are. A rigorous critique of Amnesty’s methodology would need to engage with this layer of evidence directly rather than treating the investigation as resting solely on interviews inside Afghanistan. It does not.

None of this settles the question of what Pakistani planners believed they were targeting at the moment of the strike, which is a distinct question from what the site actually was. Precision guided strikes can be aimed at accurate coordinates built on flawed intelligence, and an honest accounting of the incident would distinguish between deliberate targeting of a known medical facility and a targeting failure rooted in faulty threat assessment. Pakistan’s silence forecloses that distinction rather than resolving it in its own favour. A government confident in a genuine intelligence failure, as opposed to a deliberate strike, has strong incentive to say so, since the reputational and legal exposure of the two scenarios differs sharply. The absence of any account, four months on, is itself informative.

There is a genuine access problem in Afghanistan that deserves acknowledgment rather than dismissal. Restrictions on UN personnel operating inside the country, including limits on female staff, have long constrained what any investigating body can independently confirm through interviews on the ground. That constraint is real and worth stating plainly. But it is analytically distinct from the evidentiary chain that actually carries this case, since satellite archives, geolocation of released military footage, and broadcast forensic review do not depend on access to Afghan territory at all. Conflating the two, treating a legitimate critique of ground level access with a wholesale dismissal of imagery based verification, is where the strongest version of Pakistan’s defence collapses into a weaker one. Under international humanitarian law, the burden of demonstrating that a strike met the tests of distinction, precaution and proportionality sits with the party that carried it out, not with the investigating organisation. Silence does not discharge that burden. It simply leaves the imagery based record as the only account on file.

The Stakes Beyond One Strike

This dispute reaches well past the fate of one investigation into one strike. Pakistan is not a party to the Rome Statute, and the UN Security Council has shown no appetite for a binding inquiry into cross border strikes in Afghanistan, leaving no institution with the authority to compel either side toward resolution. Into that vacuum, open source verification, satellite archives, independent geolocation, forensic review of released military footage, has effectively become the closest available substitute for enforcement, even though it carries no legal power to compel compliance or impose consequences. How this case resolves, or fails to resolve, will help set the operating norm for a growing number of asymmetric conflicts in which a state conducts cross border strikes against a non state or quasi state administration controlling the affected population. If credible, independently corroborated evidence of civilian harm at a protected medical site produces no formal accountability, the lesson other militaries draw from this dispute will be that silence is a survivable strategy, and that the proportionality and precaution requirements of international humanitarian law can be managed through delay rather than compliance. That is the systemic risk this case actually tests, well beyond the question of what happened at one rehabilitation centre in March.

Three Paths From Here

Three trajectories are plausible over the coming months, none of them mutually exclusive in sequence.

The likeliest path, on available evidence, is continued silence accompanied by informal briefings to allied capitals disputing Amnesty’s methodology without engaging its strongest evidentiary points, particularly the independent geolocation. This preserves ambiguity domestically while avoiding a formal admission that could carry legal exposure under international humanitarian law. It is a moderately likely outcome, on the order of an even chance or somewhat better given precedent in comparable disputes involving other militaries.

A less likely but plausible path is a formal Pakistani rebuttal that engages directly with the satellite and geolocation evidence, offering an alternative account of what the imagery shows or asserting an intelligence failure rather than deliberate targeting. This would be the more legally cautious route and would represent a meaningful shift from the current posture. Call this a modest possibility, distinctly less likely than continued silence but not remote.

The least likely path in the near term is a referral to an international investigative mechanism with independent access, given both Pakistan’s incentive to avoid binding scrutiny and the Taliban administration’s own resistance to international access inside Afghanistan, which cuts against any process requiring cooperation from both sides. This remains a low probability outcome absent significant diplomatic pressure from third parties with leverage over Islamabad.

The clearest near term indicator to watch is whether Pakistan’s foreign ministry issues any substantive written response to Amnesty’s evidentiary request before the anniversary of the strike in March 2027. A specific, documented reply, even one contesting the findings, would mark a break from the pattern observed since March. Continued non response through that date would itself be a data point confirming the first scenario.

The Precedent Being Set

What is being tested in Kabul is not only one contested strike but the durability of official denial once satellite archives, geolocation researchers, and broadcast forensic review become standard evidentiary tools in conflicts fought behind restricted access. For Western policy planners, the operative signal is the widening gap between what open source verification can now establish and what official statements are willing to confirm, a gap increasingly shaping how allied governments calibrate cooperation. For investors reading South Asian risk, cross border escalation between Islamabad and Kabul remains an active, under priced variable, with legal exposure now harder for militaries to contain through silence alone. For rights and development organisations, the case is a template for how methodology itself becomes the battleground before accountability is even reached.

Silence was once a viable strategy for governments awaiting the news cycle to move on. Satellites do not move on.

Both Amnesty International and Oxfam released reports this week documenting a rise in state-backed Israeli settler violence across the occupied West Bank over the past three years. What’s driving the escalation? Al Jazeera’s Marah Rayan breaks it down.