In Oct 2026, thousands of government websites in Nepal suddenly disappeared from the internet. No nationwide cyberattack was confirmed. Authorities took websites and some online services offline as a precaution. The trigger was two critical vulnerabilities in PHP, a programming language widely used for websites. Known collectively as PHP2Shell, the weaknesses could allow attackers to run malicious code on certain affected servers without logging in. Fixes existed, but officials still needed to find affected systems and apply them.
Most websites were subsequently restored. But if a software flaw can interrupt public services even after a fix exists, how ready is Nepal for the next one?
A vulnerability is a weakness attackers might exploit; discovering one does not mean an attack occurred. Nor does a patch automatically protect every system. Someone must know where the software runs, who maintains it and whether attackers arrived before repairs.
The past 12 years show how often the same cybersecurity failures have resurfaced around the world.
In 2014, Heartbleed exposed a flaw in OpenSSL, software that helps protect information travelling across the internet. The flaw could reveal sensitive server-memory data, turning an invisible software component into a risk for services people trusted. That year, Shellshock exposed flaws in Bash, a command interpreter used by Unix-like systems. Under certain conditions, attackers could execute unauthorized commands. Even longstanding software was not necessarily safe.
In 2017, the consequences became visible. WannaCry ransomware spread internationally through a Windows networking vulnerability, encrypting files and disrupting essential operations.
Britain’s National Health Service (NHS) was among the casualties. NHS England estimated that more than 19,000 appointments would have been cancelled in total. The disruption reached hospitals and patients, showing how a technical flaw can become a public-service emergency.
Microsoft had already released a security update. But a patch cannot protect a hospital computer that has not received it.
Later in 2017, Equifax, a major American credit-reporting agency, suffered a breach affecting approximately 148m people. Attackers exploited a known Apache Struts web-application vulnerability. A fix existed, but the affected system was not secured in time.
Other incidents exposed weaknesses in technologies trusted to control access.
A flaw disclosed in 2019 in Citrix ADC and Gateway was exploited in 2020, threatening remote access. Zerologon, disclosed in 2020, could let an attacker with network access compromise a Windows domain controller managing identities. In 2021, ProxyLogon attacks exploited Microsoft Exchange email-server weaknesses, sometimes leaving attackers with persistent access.
Systems meant to connect or protect an organization could themselves become entry points.
Then came Log4Shell in Dec 2021. It received the maximum severity score of 10 out of 10. That reflected how technically serious the flaw was, not how much financial damage it caused.
Log4Shell affected Log4j, which many Java applications use to record activity. Under vulnerable conditions, specially crafted input could trigger malicious code.
Log4j could sit inside another software product much like an ingredient inside packaged food. An organization might know the product it purchased without knowing every component inside it. The risk was particularly insidious: a business could be exposed through code it had never knowingly selected.
For many, the first challenge was discovering their exposure.
The lesson applies to PHP2Shell: an office may know its website but not every component installed by its developer, host or contractor.
In 2023, MOVEit Transfer exposed another dependence. Attackers exploited file-transfer software to steal information from organizations whose suppliers used it.
A supplier’s weakness had become somebody else’s problem.
Also in 2023, CitrixBleed exposed session information in certain remote-access products. Stolen tokens could let attackers impersonate legitimate users, potentially bypassing multifactor authentication.
A patch was not necessarily the end of the response: organizations also had to investigate whether attackers had already gained access.
These are not history’s 10 worst vulnerabilities. The software changed. The recurring failures did not. A flaw becomes a crisis when software goes unnoticed, patches fail to reach systems, responsibility is unclear or attackers arrive before the fix.
Nepal has not merely watched these crises from afar. It has experienced serious cyber incidents of its own.
In 2017, WannaCry reportedly affected around 80–90 computers at a Nepali financial institution. That year, attackers compromised NIC Asia Bank’s SWIFT environment and issued fraudulent international transfer instructions. In Sept 2026, reported ransomware at a shared data centre serving 72 brokerage firms contributed to a suspension of stock-exchange trading.
The consequences ranged from infected computers and fraudulent transfer instructions to interrupted markets. Stolen credentials and other compromised data can also circulate through dark-web markets long after an incident, reinforcing the need for continued monitoring. PHP2Shell did not arrive in a country new to cyber risk. The more revealing question is whether each incident leaves Nepal better prepared for the next one.
Google’s researchers have also demonstrated AI-assisted discovery of previously unknown software vulnerabilities.
PHP2Shell brought responsibility into sharp focus. Thousands of government websites on the shared Government Integrated Website Management System (GIWMS) were patched centrally, while others depended on individual offices or vendors and took longer to restore.
Knowing who maintains a system can matter almost as much as knowing how to patch it. Clear ownership creates accountability for identifying exposure, fixing it and confirming services are safe to restore.
When a flaw threatens thousands of systems, someone must coordinate warnings and responses. India's national Computer Emergency Response Team, CERT-In, has operated since 2004. Nepal developed community initiatives, including npCERT in 2016, before establishing its government-led National Cyber Security Centre in 2024.
The roughly two-decade gap between India’s national CERT and Nepal’s government-led centre is notable, although establishment dates alone do not measure capability. What matters is whether warnings reach those who must act.
Nepal’s National Cyber Security Centre issued a PHP2Shell advisory on Oct 22. That precautionary response was encouraging, but an advisory is only the beginning: agencies must identify exposure, patch systems, check for earlier exploitation and restore services safely. Public technical accounts of some earlier incidents remain limited, although that does not mean investigations were not conducted. What matters is what happens after the warning.
Expertise must reach beyond government and Kathmandu. The Open Worldwide Application Security Project (OWASP), an international software-security community, has a Kathmandu chapter; similar networks could take root in other regional centres. Universities can build expertise through research into secure software, vulnerabilities and malware.
Electricity networks, connected vehicles and drones depend on software. So do industrial-control systems such as SCADA, which monitor and control physical processes. Cybersecurity and physical security increasingly overlap.
PHP2Shell does not prove Nepal unprepared. Its precautionary response showed capacity; the question is whether readiness extends across essential services and suppliers.
What would greater readiness look like?
Start by knowing what software and components an organization runs. Developers can use the OWASP Top 10 for web-application risks and MITRE’s CWE Top 25 for dangerous software weaknesses. Independent vulnerability assessments and penetration testing can expose problems before attackers do.
Researchers need authorized reporting channels. Clear vulnerability-disclosure programs and carefully scoped bug bounties can encourage responsible reporting.
Microsoft’s monthly Patch Tuesday illustrates why updating should be routine, not improvised during a crisis. Critical or actively exploited flaws, however, cannot always wait for the next cycle.
Fixes must be verified, and possible earlier intrusion investigated. Multifactor authentication helps against account attacks but cannot repair a flaw allowing code execution without login. Updated antivirus and endpoint protection can detect or block malicious activity, but organizations also need monitoring, tested backups and rehearsed recovery plans. Prevention will sometimes fail, so an organization must be able to detect an intrusion and recover operations rather than merely reinstall software.
Software is only one route for criminals. Sometimes exploiting trust is easier than exploiting code.
The FBI documented North Korean actors posing as recruiters and sending a malicious pre-employment test to an employee of a cryptocurrency-wallet software company. Even technical experience cannot eliminate deception.
Citizens face similar risks. Following Nepal’s Aug 2026 floods, police warned about fraudulent relief appeals, fake QR codes and phishing links. Emergencies make urgency and generosity targets. An apparently helpful donation request can therefore become the entry point for fraud.
Warnings must reach professionals and ordinary citizens quickly, in languages they understand.
PHP2Shell showed the value of precaution—and of knowing which systems exist, who maintains them and how quickly they can be secured.
The next serious flaw may emerge in software few organizations realize they use. Nepal cannot predict its name or date, but it can strengthen the expertise, institutions and operational discipline needed to respond without unnecessary disruption. The next vulnerability is inevitable. The next crisis need not be.
The author is an Ireland-based cybersecurity professional and writer