TrendAI has issued a threat advisory for organizations in Japan after its researchers found that publicly disclosed incidents during the first nine months of 2026 exceeded the total they tracked for all of 2025. September accounted for 86 incidents. These figures describe the company’s monitoring of disclosed events, rather than a comprehensive national count of every cyberattack or breach. TrendAI also explicitly states that it has not established a connection between the recent Japanese incidents and attackers’ use of AI, despite discussing AI’s potential to accelerate criminal activity.
The distinction between disclosed incidents and total activity matters because reporting is an imperfect measure. Some organizations discover problems late, some disclose little, and others never make an incident public. Changes in disclosure practices can also affect comparisons between periods. Even with those limitations, a rising count deserves attention because it represents organizations experiencing problems serious enough to become visible outside their own systems.
The business consequences extend beyond the immediate victim. Companies rely on hosting providers, software services and suppliers that may hold data or support essential operations. A breach at one participant can create costs and disruption for others, even when their own networks remain intact. Security planning therefore has to account for the dependencies through which a failure can spread, including whether critical work can continue while a provider investigates or restores service.
AI adds a plausible pressure on that environment, but plausibility should not be confused with attribution. Tools that assist with translation, reconnaissance or message generation could make it easier for attackers to operate across languages and target more organizations. That explains why defenders should prepare for changes in attacker productivity. It does not establish that AI caused a particular increase in observed incidents, and the Japan figures should not be presented as proof that it did.
For organizations assessing the warning, the most useful response is to examine exposure and recovery capacity. Public-facing applications, privileged access and supplier dependencies deserve close attention because weaknesses there can create an entry point or amplify the damage. The growing disclosed count is a reason to test those arrangements against actual operating conditions. A security plan has to work when a service fails, not only when the audit documents are complete.