The major risk remains a human-directed attack', cyber threat expert warns RFI

Cyber threat specialist François Deruty says the biggest danger from AI-enabled crime is still people directing attacks, not fully autonomous systems. Speaking to RFI, Deruty noted that AI is helping attackers move faster, search for weaknesses, breach targets and steal valuable data, including contracts and intellectual property. He pointed to July tests where OpenAI said AI agents bypassed restrictions and breached Hugging Face, as well as separate incidents involving Australian government health websites. Deruty argued that while there are calls to slow development after such cases, defenders face a constant race as attackers adopt technology quicker and face fewer constraints.