National CERT Warns Banks, Power and Water Systems of AI-Driven Cyberattacks

By Fatima Raza | November 7, 2026

Pakistan’s National Cyber Emergency Response Team (National CERT) has warned the country’s banks, financial institutions and the operators of power, water, oil and gas, manufacturing and transport systems that artificial intelligence is now making cyberattacks faster, broader and harder to stop. In an advisory dated November 5, 2026, titled “Emerging AI-Enabled Cyber Threats Targeting Financial Institutions and OT/Critical Infrastructure”, the team says AI, automated bots, botnets and agentic tools are being used to find weaknesses, steal login credentials and launch coordinated attacks at machine speed. This is a warning, not a report of an incident: no bank or utility has been named as a victim, no attacker identified, and no breach confirmed.

For anyone who banks on a phone or pays bills online, the message is straightforward: the systems behind everyday life are being probed by tools that never sleep, and banks and infrastructure operators need to raise their guard now.

What the advisory actually says

National CERT’s warning is blunt about the shift underway. Attackers are no longer relying only on a skilled hacker sitting at a keyboard. The advisory says an increase has been observed in “AI, automated bots, botnets and agentic tools to automate reconnaissance, identify vulnerabilities, conduct credential-based attacks and launch coordinated cyberattacks”. Agentic tools are software agents that carry out multi-step tasks with little human supervision, so a campaign that once needed a whole team can now be run by a handful of people.

The point is scale. AI does not invent a new kind of crime. It takes the old ones, password guessing, probing for weak spots, flooding systems with fake logins, and runs them thousands of times faster and across many targets at once.

National CERT, which operates under the Cabinet Division, said the advisory is relevant to all sectors, but singled out two areas: financial services, and operational technology and critical infrastructure. That pairing connects the money in your account to the electricity and water in your home.

Banks in the crosshairs: core systems, apps and APIs

For banks, the advisory names the systems at risk: core banking systems, the back-end platforms that hold real account records, digital banking platforms, payment systems, application programming interfaces (APIs), and customer-facing applications.

Three scenarios are spelled out for banks: disruption of online services, automated credential attacks using stolen or guessed usernames and passwords, and manipulation of the balances displayed to customers.

That last point needs careful reading: the advisory means the balance shown on your screen, not the money itself. If an attacker can make your app show a wrong balance, even briefly, the damage is panic and a collapse of trust in digital banking, even if the core records underneath are untouched. That is why banks must reconcile displayed balances against core records constantly.

The stakes are real: salaries land in accounts, school fees are paid from phones, and small businesses run on mobile transfers. A day-long disruption is a crisis for people living paycheque to paycheque.

Power, water, gas and transport: the OT warning

The second half of the advisory looks beyond banks to the physical systems that keep a country running, the operational technology, or OT, that controls valves, pumps, turbines and grids. The weaknesses are familiar ones: third-party service providers, connected internet-of-things devices, vulnerabilities in industrial control systems and insecure remote access.

The targets named are equally concrete: oil and gas installations, refineries, pipelines, water utilities, power generation and distribution networks, manufacturing plants and transport systems. A refinery that stops or a grid that trips has consequences you can feel within hours.

The defence the advisory recommends is architectural: separate OT networks from IT systems, so the computers that run a power plant do not sit on the same network as the office email system. Years of convenience have blurred that line, and redrawing it is slow, expensive work.

The European supermarket example, and an important caveat

To show what an attack on connected operational systems can do, the advisory cites recent incidents in Europe in which supermarket refrigeration and freezer systems were targeted: food spoils and shelves empty, a loss that has nothing to do with stolen data.

But the advisory adds a caveat that honest reporting must keep front and centre: “However, AI involvement in said incidents has not been confirmed yet.” The European attacks show what connected-system attacks can do, but they do not prove an AI-driven campaign.

What banks have been told to watch for

The advisory also gives banks a practical checklist of warning signs: automated login attempts, repeated authentication failures, unusual API requests, changes in application configurations and abnormal transaction patterns. Each sign is small on its own, but together, in volume, they form the fingerprint of an automated campaign.

One honest limitation: the advisory carries no binding enforcement power and no budget. It tells institutions what to do, but cannot compel them to do it. Whether the warning turns into real change depends on each bank and utility acting on it, and on the State Bank’s own rules for core banking and network security, which do carry mandatory audit and reporting requirements.

Why this warning lands now

The advisory arrives in a week when the threat is visible on both sides of the border. India’s CERT-In issued its own advisory on November 4 about an AI-assisted campaign it calls “CRAFTY FALCON”, targeting banks, non-banking finance companies, fintech firms and payment providers. The two advisories are separate, but they describe the same shift: AI is lowering the skill and effort needed to run a serious financial cyberattack.

Pakistan’s own exposure is already documented. Only three days before the advisory, National CERT Director General Haider Abbas told the National Assembly’s IT committee that Pakistan has faced more than 1,000 cyberattacks so far in 2026. This is not a hypothetical threat: the country is already being hit, and the advisory says the hits are about to get smarter and faster.

There is also a capacity question. According to reporting of the hearing, National CERT runs on a staff of around 70 people, with about 70 per cent of its threat-monitoring work still done manually. That detail comes from a single roundup of the hearing, so it should be read as an indication rather than a confirmed figure, but it underlines the challenge: a small, largely manual team warning a whole financial system about machine-speed attacks.

What customers should do

For ordinary bank customers, the advisory is not a reason to panic or to stop using digital banking. It is a reason to do the few things that genuinely reduce risk: turn on multi-factor authentication on your banking app and email, use a strong, unique password for your bank, and check your balance regularly against the SMS and email alerts your bank sends. If the numbers do not match, call your bank straight away.

Be suspicious of urgent messages asking you to click a link or share a one-time code; banks do not ask for those by phone or text. If your app stops working or shows a wrong balance, report it at once, and the Banking Mohtasib’s record of securing refunds for customers shows complaints, properly made, do get results.

The bigger picture is that Pakistan’s digital economy depends on trust in systems most people never see. The tools exist: multi-factor authentication, network separation, balance reconciliation, constant monitoring. What matters now is whether every bank and utility actually uses them. As the financial system keeps digitising, and as privatisation puts more state-owned entities into private hands, the attack surface will only grow. The advisory has done its part by naming the risk. The rest is execution.