Prompt Engineering in CTOC Instruction: Countering Transnational Organized Crime with Artificial Intelligence

Cüneyt Gürer, “Countering Transnational Organized Crime with Artificial Intelligence,” The Clocktower Series GCMC CT25-04 (George C. Marshall European Center for Security Studies, December 2025).

This December 2025 Clocktower Series concept paper presents a practitioner-oriented “CTOC with AI” model piloted at the George C. Marshall European Center. Author Dr. Cüneyt Gürer argues that integrating large language models into transnational organized crime strategy is now essential. The paper advocates a “+AI approach”—using AI to augment rather than replace human expertise.

As Gürer explains, “prompt engineering is the art and science of crafting precise and effective inputs to guide AI models toward relevant and accurate outputs.” The methodology follows six steps: data collection, AI-assisted analysis, policy options generation, risk identification, human validation, and final recommendations. Critical to this framework is keeping “humans in the loop, ensuring AI remains a tool serving human objectives rather than acting autonomously.” The model addresses core ethical considerations including bias, privacy, and transparency.

Gürer emphasizes that AI can “analyze complex data, generate policy options, and develop scenarios under human supervision,” but only with proper safeguards. The urgency is clear: criminal networks already use AI to enhance operations. The paper concludes that “integrating AI into CTOC work is not optional but an essential professional requirement for the workforce of the future.”

Integrating AI into CTOC Strategy and Policymaking: A Step-by-Step Approach

The CTOC with AI model provides a structured methodology to integrate AI into policy and strategy development processes. This systematic approach guides decision-makers through the entire analysis and policy formulation cycle.

Step 1: Data Collection and Entry. The initial phase involves collecting relevant data and entering this data into an AI platform.

Step 2: Data Analysis. Once the data is entered, the AI processes and analyzes the content. Critically, this phase relies on the effective application of prompt engineering techniques to guide the AI’s analytical focus including extracting specific insights and shaping desired
outputs.

Step 3: Policy Options Generation. Following data analysis, the AI generates various policy options and recommendations. This step represents a crucial transition from raw data and analytical findings to actionable insights and potential strategic responses for countering TOC challenges.

Step 4: Risk Identification and Mitigation. An essential and integral part of the process, this step involves explicitly identifying potential areas where bias or hallucinations might occur in the AI’s output.

Step 5: Results Validation. Human experts conduct a thorough review of the AI-generated results. This validation step is essential to ensure that the outputs are logical, make sense within the real-world context of TOC, and align with broader strategic objectives.

Step 6: Findings and Policy Recommendations. The final step involves synthesizing the AI-driven analysis with human validation. This culminates in a comprehensive summary of findings and the formulation of concrete, actionable policy recommendations designed to address the identified transnational organized crime challenges.