With phishing attacks becoming more sophisticated, leveraging AI for cybersecurity defence has become essential. Traditional rule-based detection methods are no longer sufficient. Instead, AI-driven solutions that understand phishing tactics, identify deception patterns, and adapt to new threats are necessary to protect individuals and organisations. From April 24 to 26, 2025, COcyber’s AI Cybersecurity Deephack brought together cyber enthusiasts from academia, research and the start-up scene for an intense 48-hour challenge to answer one critical question: can artificial intelligence stop the next big phishing attack? During the DeepHack, participants used natural language processing (NLP), anomaly detection, and existing phishing threat intelligence sources to develop AI-driven systems able to analyse linguistic anomalies, behavioural patterns, and suspicious URLs in emails and websites to detect, identify and prevent phishing attacks in real-time. The challenge required participants to focus on dual-use cybersecurity technologies to ensure that the final solutions could target both civilian and defence cybersecurity applications, while addressing practical deployment issues such as multilingual attacks, AI bias, infrastructure constraints, and adversarial threats.