Two centuries ago, a British social reformer named Jeremy Bentham envisioned a model prison in which every prisoner’s cell was under constant surveillance from a centralised platform. The idea being that if a person believed they were under constant surveillance, they wouldn’t reoffend and eventually reform the habits which led to crime in the first place. The name for this prison was the Panopticon. Today, Bentham’s vision of the panopticon has become reality, not in prison yards and poorhouses, but on the streets, in the form of an ever-growing network of surveillance and security cameras, driven by AI and predictive policing algorithms in pursuit of ever-greater reductions in crime. However, these new technologies were introduced with the intention of stopping crime. The fundamental philosophy governing their implementation lacks oversight and accountability, posing a major risk of government overreach and private abuses, and endangering civil liberties and police accountability.
The logic of the panopticon is a tempting one for governments and law enforcement. If you design an environment properly, it might be possible to eliminate crime. However, the cost of this utopia is securing acceptance and compliance from individuals in the new environment. You can do this through force, which is inefficient, in specialised environments, which is expensive and impractical, or through complete visibility (i.e., surveillance and corrective action). The benefit of mass surveillance is that once participants come to expect mass surveillance and swift punishment, the system becomes self-regulating. People act correctly and avoid wrongdoing because they know they are always being watched and will be swiftly punished for transgressions. The cost of this system, however, is unaccountability, the dampening of civil liberties, and the over-enforcement of small, inconsequential crimes.
The proliferation of surveillance cameras and the integration of AI into surveillance networks have allowed the Panopticon logic to go from mere theory to daily practice in law enforcement. Flock is a great example of the modern panoptic surveillance paradigm. Its network of license plate-reading cameras, powered by AI, was created to help police catch criminals such as car thieves and kidnappers. By continuously monitoring and storing information about motor vehicles on public highways in a searchable database for law enforcement to use to track suspect vehicles. And has seen a modicum of success in reducing crime, with the Berkley Police Department made 52 arrests in 2025 for a range of crimes from homicide to robbery, utilising Flock systems. This system has also been used to identify and surveil protestors, target illegal immigrants, and even women receiving abortions. Leading to substantial concerns over privacy rights, abuse, accountability and consent of the governed.
These concerns aren’t reserved just for the citizens in the U.S., police officers in Germany and Switzerland have also expressed concerns that the operation of commercial predictive policing and surveillance systems as private black boxes eliminates police officers’ discretion and undermines democratic accountability. These are not insubstantial concerns, as digital surveillance and AI-enabled analysis have enabled authoritarian regimes to cheaply and effectively surveil their populations to avoid and deter dissent. For example, in China, a system of mass surveillance is analysed using AI, not only to address crime but also to identify dissidents and create social credit scores that govern access to public services. And just like in the Panopticon, once citizens learn to assume that all those things are happening, they alter their behaviour without the regime having to resort to physical repression.
Other countries across Asia, Africa, and Latin America have also begun adopting similar systems to better control their diverse populations and to avoid popular uprisings and dissent, such as those of the Arab Spring in 2011. This is best seen in the push by Chinese companies like Huawei to develop so-called “Safe Cities” projects across Southeast Asia, Africa and the Middle East. The Asia-Pacific region is particularly exposed to these quick-fix safe city solutions, as its governments tend to be semi-authoritarian and have limited administrative capacity, creating the perfect storm of limited political oversight coupled with financial and political incentives to implement these systems.
Putting aside governmental abuse, these systems of universal AI-driven surveillance also put the general public at risk from foreign and domestic bad actors. Foreign adversaries could gain access to these networks and then use them to track dissidents living abroad, or government officials and their families. Then use the information gathered to pressure, blackmail or threaten them into compliance with their wishes. While criminals could mine the data gathered by these systems for sale to corporations or on the black market. Circling back to Flock, it has been revealed that it is using unvetted overseas gig workers to train its AI. This includes specific instructions to categorise vehicles by make and colour, transcribe licence plates, label pictures of people, and label noises picked up by the software. An unprecedented invasion of privacy and putting the U.S. public data and personal movements at risk of exposure to foreign adversaries and bad actors
Thus, while panoptic systems are sold as a way to more efficiently use police resources and more effectively solve crime. Their potential to suppress dissent and enforce compliance, combined with the current lack of a clear regulatory framework, makes these systems pose an existential risk to civil liberties and democratic freedoms. While also exposing domestic populations’ daily movements to foreign adversaries and criminal elements. If we are to capitalise on the great potential these systems offer for crime-fighting and police accountability, while preserving civil liberties and democratic freedoms. The design and operation of these panoptic systems must be made more accountable and transparent, with tighter safeguards for the privacy and rights of everyday people.
Recommended Readings
Brayne (2020) Predict and Surveil
Leese (2022) Staying in Control of Technology: Predictive Policing, Democracy, and Digital Sovereignty
Taylor et al. (2020) The Digital Dictators: How Technology Strengthens Autocracy