Millions of personal records in Japan and South Korea have been exposed, The Straits Times reports, as both countries absorb a surge of cyberattacks that weaponise artificial intelligence.

The headline number is blunt: millions of individual records, across two of Asia's most wired economies, now sit outside the perimeter. The culprit is not a novel zero-day or a lone insider. It is a measurable uptick in AI-powered attacks — offensives that automate reconnaissance, phishing and credential stuffing at a scale human operators cannot match.

What does this change for CISOs in the region?

It moves the assumption set. Attackers no longer need teams; they need tooling. AI has cut the cost of mounting a credible campaign against Japanese and Korean enterprises, which means the frequency of attempts rises even where target sophistication does not. Volume, not novelty, is doing the damage.

Personal records are the payload of choice because they resell. A breach of this scale in two markets with strict data-protection regimes — Japan's APPI and South Korea's PIPA — also carries regulatory consequences that outlast the incident itself.

Why AI changes the economics of the attack

Automation compresses the kill chain. What once took an operator days — target selection, lure crafting, follow-up phishing in credible local languages — now takes minutes. Defenders running manual triage lose that arithmetic contest every time.

The practical response is not exotic. It is:

• Rate-limiting and anomaly detection tuned for machine-speed probing
• Credential hygiene that assumes phishing succeeds at volume
• Data minimisation, so a breached record store is a smaller one

What to watch next

Attribution and breach-notification detail will surface through official channels in the coming weeks. For security leaders in both markets, the question is no longer whether AI-assisted attacks arrive. It is whether detection and response cycles are fast enough to matter when they do.

Millions of exposed records suggest, so far, they are not.