Artificial intelligence is increasingly integral in combating cybercrime, with law enforcement agencies like the gendarmerie leveraging AI tools to enhance investigations. General Patrick Perrot discussed innovations such as Odip, which identifies child pornography images while safeguarding investigators from trauma, and AUTHENTIK IA, aimed at addressing deepfake threats. The gendarmerie is also utilizing platforms like Ollama for language models to aid administrative tasks, while recognizing both the potential and challenges posed by AI in cybersecurity.
In the realm of cybersecurity, artificial intelligence (AI) has become both a crucial tool for law enforcement and an attractive asset for cybercriminals. As the AI Summit kicks off on February 10, many are still trying to understand the vast potential of this technology. However, law enforcement agencies, like the gendarmerie, are already harnessing AI to enhance their crime-fighting capabilities. General Patrick Perrot, the AI coordinator for the gendarmerie, highlighted the integration of AI into investigative processes during the Cybershow Paris 2025.
How AI is Revolutionizing Law Enforcement
General Perrot: Our foremost objective is to combat cybercrime, and AI serves as an invaluable asset in this endeavor. We are currently in the developmental phase of implementing various AI-supported tools.
One of the key innovations being developed is Odip, a tool designed to identify child pornography images. This AI-driven solution accelerates the analysis of sensitive materials while protecting investigators from direct exposure, thereby mitigating trauma. Odip classifies images without displaying them, transforming them into actionable data for law enforcement officials. As of now, the tool is in its trial stages, boasting a 4% false negative rate.
Another significant project is “AUTHENTIK IA,” which targets the rising threat of deepfakes. The advancement of this tool requires time and ongoing development, as the tactics employed by criminals are constantly evolving, impacting the effectiveness of our countermeasures.
Developing AI Tools for Enhanced Investigations
General Perrot: We primarily utilize Ollama, an AI platform that enables the execution of language models directly on computers without needing an Internet connection. We also enhance our tools with solutions sourced from French and European developers. While the ideal scenario would involve a fully French-developed tool, it’s crucial for us to progress without delay; waiting for complete finalization could hinder ongoing investigations.
Our main focus is to deploy AI to support our investigative efforts, and we are simultaneously exploring language models that can assist gendarmes with human resource tasks and streamline document management. AI is perfectly suited for these administrative functions, and there’s no reason we shouldn’t leverage its capabilities.
Addressing Concerns About AI’s Impact
General Perrot: I maintain a balanced view of AI. Yes, it has vulnerabilities, and we observe its misuse by cybercriminals through the mass production of deepfakes, automation of fraudulent communications, and even the proliferation of malicious chatbots on illicit platforms. The list of threats is extensive.
However, it’s important to recognize that AI has the ability to identify its own weaknesses, anticipate issues, and alert us accordingly. Furthermore, it is poised to optimize various sectors, including administration, investigation, environmental research, and more. I firmly believe that the advantages of AI will ultimately benefit society as a whole.