Artificial Intelligence is no longer a futuristic idea. It is already woven into daily life even in Pakistan. TikTok filters, predictive text on phones, banking chatbots and facial recognition at airports all show how quietly it has arrived. The problem is that while the technology is moving quickly, the rules are missing. And when powerful tools spread without safeguards, they bring not only innovation but also the risk of abuse.
At the moment Pakistan is simply observing how others act. Europe has passed the EU AI Act, the first comprehensive law that classifies AI systems according to their level of risk. The USA has leaned on voluntary standards, allowing industry to regulate itself. China has chosen the opposite route, embedding AI into its system of surveillance and control.
These examples may tempt Pakistan to copy them, but they cannot be transplanted here. Each country writes its rules according to its own priorities. Europe emphasizes privacy and human rights. China emphasizes state stability. Pakistan faces weak institutions, low digital literacy and mistrust between citizens and government. Imported rules will not fit these conditions.
The temptation will be to wait, as Pakistan often does, until problems erupt. But AI is not like a fuel shortage or a flood. Once embedded in everyday systems, it becomes invisible and irreversible. If boundaries are not set now, we will be governed by technologies that we neither understand nor control. The countries that will succeed in the AI era are not only those inventing the smartest tools. They are also the ones writing the smartest rules. Pakistan cannot afford to remain a bystander. If we do not shape our own AI story, others will shape it for us and the ending may not be one we like
The risks of waiting are serious. During an election, deepfake videos could flood WhatsApp groups and spread lies faster than facts can correct them. AI-powered surveillance could quietly expand in cities without oversight, justified in the name of security. Private companies might use facial recognition for advertising without citizens even realizing their images are being collected. These scenarios are not distant speculation. The USA already struggles with AI-generated deepfake videos, a devastating tool of online harassment targeting women. China’s widespread facial recognition has drawn global alarm. In Kenya, AI-powered credit apps have pushed people into cycles of debt. Pakistan will face similar challenges if it does not prepare.
Pakistan’s vulnerabilities are sharper than most. Disinformation already poisons politics, with doctored images and edited audio circulating widely. AI-generated deepfakes would make this even worse. Banks and telecom companies are beginning to use AI for customer service and financial decisions. If an algorithm wrongly blocks a SIM card or denies a loan, who is responsible? The company, the software developer or the system itself? In Pakistan’s weak regulatory culture, citizens will be left helpless. Language is another barrier. Most AI tools operate in English, excluding Pakistanis who rely on Urdu or regional languages. No doubt, surveillance is an additional risk. With already implemented biometric databases and Safe City projects already in place, AI could rapidly expand monitoring powers. Without legal protections this would create unchecked authority.
The lesson from abroad is to adapt rather than imitate. Europe’s insistence on transparency, requiring companies to explain how AI affects people’s rights, provides a model that Pakistan could apply in banking, policing and healthcare. The American approach shows how voluntary standards can encourage innovation but also invite abuse. That is a mistake Pakistan cannot afford. China shows the efficiency of state-led AI but also the cost, which is the erosion of freedom. Pakistan, still struggling to strengthen its democracy, must avoid this path.
A Pakistani approach should begin with the most urgent issue, elections. If AI-generated disinformation is left unchecked, democratic credibility will be the first casualty. The Election Commission, supported by technology platforms, should ban deepfakes in campaigns and create rapid response systems to detect and remove them. Consumer protection must also be a priority. Citizens using AI systems in banks, hospitals or telecom services should be informed and have the right to request a human review if a decision feels unfair. Biometric data must be shielded by strict legal protections to prevent misuse for political or commercial purposes.
Inclusion is equally important. If AI tools for education or public services remain in English, the gap between elites and the majority will widen. Urdu and regional languages must be built into these systems from the start. Finally, Pakistan requires an independent AI commission with real authority, not a symbolic office inside a ministry. This commission should bring together technologists, ethicists, civil society and policymakers to ensure AI is developed and used responsibly.
The temptation will be to wait, as Pakistan often does, until problems erupt. But AI is not like a fuel shortage or a flood. Once embedded in everyday systems, it becomes invisible and irreversible. If boundaries are not set now, we will be governed by technologies that we neither understand nor control. The countries that will succeed in the AI era are not only those inventing the smartest tools. They are also the ones writing the smartest rules. Pakistan cannot afford to remain a bystander. If we do not shape our own AI story, others will shape it for us and the ending may not be one we like.
The writer is serving as Director of the Institute of Humanities and Arts at Khwaja Fareed University of Engineering and Information Technology, Rahim Yar Khan, Pakistan.
Allah hu Akbar