Ouagadougou: Mathematics constitutes an essential lever for the development and security of artificial intelligence (AI) systems, said Professor Stanislas Ouaro, Director of the Laboratory of Mathematics and Computer Science of Joseph Ki-Zerbo University (LAMI), during a conference organized by his Laboratory. According to Burkina Information Agency, the event, organized in partnership with the Mathematics Students' Club, focused on the mathematical aspects of artificial intelligence security, following two previous sessions on quantum technologies and post-quantum cryptography. Professor Ouaro highlighted the pivotal role of mathematics in AI, especially through optimization techniques that bridge gaps between computer model predictions and reality. He explained that mathematical advances, particularly the gradient descent algorithm, have significantly contributed to the rise of artificial intelligence. The development of AI is not the only focus; Professor Ouaro emphasized mathematics' role in bolstering cybersecurity against hacking, data theft, or manipulation of AI systems. He noted the importance of developing national competencies in these areas through international scientific collaborations. Furthermore, Professor Ouaro observed that computer science students often lack a strong mathematical foundation, which he believes limits their research depth. He urged them to enhance their mathematical knowledge to better master AI technologies. The conference addressed issues concerning AI system protection against cyberattacks and data theft. He cited a 2024 fraud incident in Hong Kong, where approximately $25 million was stolen using AI tools, as an example of the challenges faced. In conclusion, Professor Ouaro likened cybersecurity to a continuous race between "the thief and the policeman," asserting that protection strategies must evolve with new attack techniques, with mathematics serving as the foundation for this evolution. The conference was hosted via videoconference from Montreal, Canada, by Dr. Pi erre Sogo Sanon, under the theme: "Mathematical and cryptographic aspects of AI security and protection, including adversarial learning, model robustness, homomorphic encryption, differential confidentiality."