IBM Japan has released the Japanese edition of its “2025 Cost of a Data Breach Report.” The study highlights a growing gap between AI adoption and the maturity of AI security and governance. While only a small percentage of surveyed organizations reported AI-related breaches, this first assessment of AI security, governance, and access controls indicates that AI has already become an attractive and vulnerable target.
With enterprise AI adoption accelerating, this year’s study for the first time examines AI security and governance, data targeted in AI-related incidents, the costs of AI-driven breaches, and the risks of shadow AI. Findings suggest many organizations are rushing to adopt AI without addressing security and governance, leaving unguided systems more vulnerable and breaches more expensive.
“Our findings demonstrate that a gap already exists between AI adoption and oversight, and that threat actors are beginning to exploit it,”
said Suja Viswesan, Vice President, IBM Security and Runtime Products.
“This research exposes a lack of basic access controls in AI systems, leaving sensitive data exposed and models vulnerable to manipulation. As AI becomes more deeply integrated into business operations, AI security must be treated as a foundational element. Failure to act will result in not just financial costs, but also a loss of trust, transparency and control.”
The study revealed that organizations leveraging AI and automation extensively in their security operations reduced breach costs by an average of $1.9 million and cut the breach lifecycle by about 80 days.