Governments and law enforcement must not race to adopt AI for security without first ensuring its governance is worthy of trust.

The Role of Artificial Intelligence in Policing — Ethical Boundaries and Legal Frameworks

by ChatGPT-4o

The presentation titled "The Role of Artificial Intelligence in Policing" by Dr. Camilla Della Giustina (JD) explores a timely and critical question: how can law enforcement agencies responsibly integrate AI technologies—especially Automated Facial Recognition (AFR)—without compromising democratic values, human rights, or the rule of law? This essay examines the key ideas, highlighting the most surprising, controversial, and valuable insights, and concludes with a set of policy recommendations for governments and law enforcement agencies worldwide.

Most Surprising Statements

1. AI as Both “Medicine” and “Poison”
   The characterization of AI as both a cure and a toxin is a striking metaphor. This duality emphasizes the transformative potential of AI in improving public safety, while simultaneously warning of its dangers—such as eroding privacy, automating bias, and increasing state surveillance powers.

2. Proposal to Maximize AI’s Positive Elements While Maintaining Rule of Law
   While many regulatory discussions focus on curbing AI’s harms, this report proposes enhancing AI’s strategic role in policing, provided it operates within ethical and legal guardrails. This framing is optimistic and highlights the possibility of productive coexistence between public security and civil liberties.

3. Interdisciplinary Methodology Including Engineers and Sociologists
   The emphasis on a cross-disciplinary approach—including not just lawyers, but engineers, informatics experts, and sociologists—is notable. This reflects a growing recognition that AI regulation cannot be tackled by lawyers or policymakers alone.

Most Controversial Statements

1. Mass Video Surveillance and Biometric Collection as “Control Power”
   The paper introduces a legal-philosophical idea of AI reinforcing the “control power” of the state—particularly through mass surveillance and facial recognition. This framing aligns with concerns about authoritarian drift and the normalization of panoptic governance, especially when surveillance occurs in public spaces where the right to privacy is legally ambiguous.

2. Use of Case Law to Justify AI in Policing
   The presentation references high-profile legal decisions like Catt v UK, Bridges v South Wales Police, and Big Brother Watch v UK to contextualize the debate. However, it somewhat controversially uses these cases to propose a legal framework rather than cautionary boundaries. For instance, Bridges revealed critical legal shortcomings in the deployment of AFR systems, yet the presentation pivots toward refining legality rather than questioning legitimacy.

3. “Strategic Importance” of AI in Law Enforcement
   The emphasis on AI’s strategic importance to policing risks prioritizing security over rights. While it’s pragmatic, such framing could embolden unchecked surveillance initiatives under the guise of innovation, particularly in jurisdictions with weak democratic institutions.

Most Valuable Insights

1. Balancing Test: Public Security vs. Human Rights
   The idea of a “balancing test” echoes core jurisprudential principles under the European Convention on Human Rights (ECHR). It calls for a proportional approach—using AI only where necessary, and only when its benefits outweigh the risks to individual freedoms.

2. Need to Enhance Citizen Awareness and Legal Transparency
   A vital insight is the proposal to improve transparency and citizen awareness. Ethical deployment of AI must go hand-in-hand with public education, democratic oversight, and accountability—particularly given AI’s black-box nature and potential for abuse.

3. Reference to European Data Protection Board Guidelines (2023)
   Citing the EDPB Guidelines on facial recognition lends the report both authority and urgency. These guidelines underscore that biometric surveillance, even for law enforcement, must remain the exception, not the rule. Their inclusion reinforces the importance of aligning national practices with evolving European norms.

Points of Agreement

• Ethical Framework Is Essential: I fully agree with the emphasis on a human-centric, ethical framework that prioritizes legality and proportionality. AI must not be allowed to erode civil liberties or operate outside legal scrutiny.

• Interdisciplinary Approach: Policing AI is not just a legal issue—it’s also a technical, social, and psychological one. Involving multiple disciplines ensures more holistic policy development.

• Transparency and Accountability: Ensuring transparent decision-making and explainable AI is crucial. Opaque AI systems in the hands of police can lead to unreviewable injustices, especially against marginalized communities.

Recommendations for Governments and Law Enforcement Agencies Worldwide

1. Codify AI Governance Principles into Law
   National legislatures should enact clear laws governing the permissible use of AI in policing. These laws must uphold the principles of necessity, proportionality, and legality—ensuring AI systems are used only where human rights can be demonstrably protected.

2. Ban or Strictly Limit Mass Surveillance via Facial Recognition
   Following the precedent of cities like San Francisco, governments should consider outright bans—or at minimum, strict limitations—on AFR in public spaces. Passive, real-time biometric surveillance is incompatible with a free society unless accompanied by judicial oversight and temporal/spatial limitations.

3. Establish Independent AI Oversight Bodies
   Create or empower independent regulatory agencies to audit police use of AI tools. These bodies must be legally empowered to investigate misuse, issue fines, and suspend technologies when violations occur.

4. Mandate Human-in-the-Loop Decision-Making
   Ensure that AI recommendations—whether in predictive policing, surveillance analysis, or identification—do not replace human judgment. Final decisions must be made by accountable humans who can be questioned and challenged in court.

5. Prioritize Training and Interdisciplinary Education
   Law enforcement agencies must train personnel not only in technical tools, but also in ethics, data protection, bias recognition, and human rights law. AI deployment must be accompanied by a shift in institutional culture.

6. Facilitate Public Participation and Transparency
   Governments must ensure the public is informed and consulted when new AI tools are introduced in policing. Regular transparency reports and participatory town halls should be mandated.

7. Embed AI Use Within International Human Rights Frameworks
   AI in policing must always comply with the standards of international treaties such as the ECHR and ICCPR. This is especially critical for countries in regions with weak institutional checks.

Conclusion

Dr. Della Giustina’s report provides a well-structured, legalistic, and interdisciplinary foundation for the discussion around AI in policing. Its most valuable contribution is its attempt to find equilibrium between the promises and perils of law enforcement automation. While it could go further in addressing the structural risks of surveillance capitalism and authoritarian misuse, it nonetheless lays out a roadmap for ethical integration.

The future of AI in policing depends not just on what can be done, but on what shouldbe done—and this, in turn, requires legal courage, public engagement, and a steadfast commitment to democratic values. Governments and law enforcement must not race to adopt AI for security without first ensuring its governance is worthy of trust.