日立 says attackers use AI; IBM sees no trace

The debate over AI's role in cyberattacks is complicated by a data problem that vendors freely acknowledge. Hitachi declined to disclose individual case counts, Accenture said Japan lacks its own statistical dataset, and IBM noted that published incidents cannot be reliably dated to the month they appear. An unnamed vendor added that reports after September exceeded 70, but some breaches dated back to July and August, inflating the apparent count.

On the defensive side, the four vendors converged on a theme that is decidedly unglamorous: understanding what must not stop, minimizing what is exposed, applying patches and preparing for the assumption of intrusion. Hitachi's Gushi called the combination of AI and human expertise the key to next-generation defense, and argued that defenders cannot keep up without AI. The unnamed vendor framed the constraint as managerial rather than technical, warning that if decisions to act or halt systems do not run smoothly, AI defenses go to waste. Accenture's Fujii said executives increasingly ask whether to prioritize hardening or resilience, suggesting those trade-offs are becoming a board-level question.