Reporting snapshot · 21 November 2026 (Asia/Karachi). The figures are detections by Kaspersky's own security products, not a count of every attack or victim in Pakistan, and the reports do not publish the underlying data set or name any affected organisation. The two newspapers describe the 6.3 million component differently, and the reports do not say whether the seven APT groups were targeting Pakistan, operating from it, or both.
Kaspersky’s security tools recorded more than 9.4 million cyberattacks in Pakistan over the year to October 2026, including a 50 per cent rise in backdoor detections, according to threat data the company presented at the Cyber Threat Intelligence Summit 2026 in Islamabad. The News reported on 21 November that the total covers November 2025 through October 2026 and comprises more than 6.3 million on-device attacks and 3.1 million web-based detections, with backdoors climbing to more than 233,000. The Express Tribune carried the same 9.4 million total and the same 50 per cent increase.
The headline number measures what one vendor’s software blocked on the devices and connections it protects. It is not a census of attacks on Pakistan, a count of victims or an estimate of financial loss. Kaspersky says 24 per cent of its users in Pakistan encountered an on-device threat and 15 per cent faced phishing or other web-based risks. Those percentages describe Kaspersky’s customer base, not the country as a whole.
What the 9.4 million counts
The two newspapers divide the total slightly differently. The News describes the 6.3 million as on-device attacks, spread largely through USB drives, CDs and encrypted installers, and adds 3.1 million web detections to reach 9.4 million. The Express Tribune describes the 6.3 million as devices targeted. The News’s breakdown is arithmetically consistent with the total, and this article treats the figure as attacks.
Software exploits remained a large category: more than 448,000 blocked attempts, down 8 per cent. That decline is not necessarily good news. It may reflect fewer attempts, better patching, or a shift towards methods that do not need a software exploit. The sharpest movement is in backdoors, which rose by half even as exploit attempts fell.
A backdoor is a way into a system that the owner did not intend to leave open. Once installed, it lets an attacker return remotely, run commands, move files and reach other machines on the same network. Unlike ransomware, which announces itself by locking files, a backdoor stays quiet. That makes it useful for stealing data, watching a network or preparing a larger operation, and harder to count from outside.
Why attackers are using tools that already work
The report’s most practical warning is about familiarity. Kaspersky says the advanced groups it tracked increasingly use legitimate administrative utilities already installed on corporate networks rather than custom malware that trips antivirus alerts. By behaving like a system administrator, an intruder can move between machines, harvest credentials and look for sensitive files without standing out.
Globally, Kaspersky recorded more than 2 million attempts to deploy dual-use hacking tools between April and September 2026. The most frequently intercepted were HackTool.Cobalt (136,084 detections), HackTool.NetExec (129,560) and HackTool.SQLMap (107,697). These are not malicious in themselves; security teams and attackers can both use them, which is why blocking them requires context rather than a list of banned file names.
Supply chains carry a similar problem. Kaspersky says nearly one in three organisations globally experienced a supply-chain attack over the past year, and The News noted a high-profile 2026 breach involving the widely used JavaScript library Axios. A supplier, contractor or software update can become the way into an organisation that would otherwise be difficult to reach. For Pakistani banks, telecom operators and government contractors, security does not stop at their own firewall.
Seven APT groups, with little public attribution
Kaspersky says it tracked seven advanced persistent threat groups in Pakistan and named two internationally active outfits, Mirage Kitten and SideWinder. APT is a loose label for groups that pursue access over time, often with more resources and patience than ordinary criminals. The names are Kaspersky’s tracking labels, not legal findings about who is behind the groups or which government, if any, directs them.
That gap matters. The News says the groups were “operating in Pakistan”, which could mean they targeted Pakistani organisations, worked from Pakistani infrastructure, or both. The reports do not name any victims, the sectors affected or the evidence behind each attribution. The seven-group figure is therefore a measure of activity Kaspersky observed, not a public finding that seven foreign intelligence operations are running inside the country.
AI is lowering the cost of the first move
Kaspersky’s researchers also pointed to artificial intelligence as a force multiplier. “Recently, threat actors have been more actively exploiting trusted relationships, legitimate tools, and emerging technologies such as AI to advance their attacks,” Yossef Abdelmonem, a senior security researcher with Kaspersky’s Global Research and Analysis Team, said, according to The News. The warning says AI can help attackers write, modify and scale malicious code faster, while the spread of autonomous AI agents inside workplaces creates new routes for accidental data leaks.
That warning points in the same direction as official advice in Pakistan. The National Cyber Emergency Response Team warned this month that AI, bots and agentic tools are making attacks on banks and critical infrastructure faster and harder to detect, as KhabarWire reported. In March, the national CERT warned that hostile actors could exploit software supply chains, as earlier reporting noted.
The two warnings are different kinds of evidence. NCERT described risks and recommended controls; Kaspersky describes what its products detected. Neither proves that a specific Pakistani bank, utility or government system was compromised. Together they point in the same direction: more covert access, more use of trusted tools, and more attacks that begin through a supplier or an employee rather than a direct assault on a perimeter.
What the numbers mean for Pakistan’s digital economy
Pakistan’s exposure is growing with its digital economy. Banks, telecom operators, e-commerce platforms and software exporters all hold data or run services that depend on networks, and the same connectivity that supports them gives attackers more places to enter. KhabarWire reported this month that IT services exports reached a record $453 million in October, taking the four-month total for the fiscal year to $1.73 billion, as the trade figures showed.
Those numbers make cybersecurity a commercial issue, not only a policing one. A supply-chain breach or a long-running backdoor can expose client data, interrupt payments and damage the reputation of firms that depend on foreign clients. Kaspersky’s data does not quantify that loss. The report also does not say how many detections were on corporate networks rather than consumer devices, the distinction that matters most for the export sector.
For individuals, the risks are familiar: phishing pages, malicious installers and infected removable media. The News reported that much of the on-device activity arrived through USB drives, CDs and encrypted installers, methods that still work because they rely on a person opening something. The 15 per cent web-risk figure points to the other common route: a link or fake login page that harvests credentials.
What would change the picture
The next signal is not another vendor total. It is whether a Pakistani institution discloses an incident, whether NCERT or the National Cyber Crime Investigation Agency publishes incident data, and whether Kaspersky releases its methodology. The figures published so far do not answer those questions.
The controls that follow are not exotic. Patching known flaws, requiring multi-factor authentication, restricting the administrative tools attackers abuse, monitoring unusual remote access and checking suppliers all address the routes the data describes. Kaspersky’s broader point is that static defences are no longer enough against groups that adapt, a claim that is difficult to test from the published figures but consistent with the shift the numbers show.
The 9.4 million figure will travel further than its caveats. It is a useful indicator of how much malicious activity Kaspersky’s sensors see in Pakistan, and a reminder that the quiet intrusion is a growing part of the threat. It is not a measure of how safe the country is.


