GCHQ's NCSC warns of 'realistic possibility' AI will help state-backed malware evade detection

The UK NCSC projects that by 2025 AI will meaningfully enhance attacker capabilities—accelerating vulnerability discovery, reconnaissance, phishing quality, and post-breach data triage—benefiting well-resourced state actors first and increasingly cybercriminals and ransomware groups; while not immediately transforming the risk landscape, AI is expected to increase the volume and impact of attacks, prompting calls for secure-by-design AI and stronger cyber hygiene, alongside policy efforts like the Bletchley Declaration and AI testing plans to manage associated risks.