As Japanese companies rush to adopt AI, governance is failing to keep pace and security risks are emerging. Tenable's latest report finds that 52% of non-human identities have excessive privileges.
Japanese companies are rushing to adopt AI to gain first-mover advantages, but their focus on implementation speed is putting governance on the back burner and creating openings for attackers.
According to Tenable's Cloud and AI Security Risk Report 2026, 52% of non-human identities, including machine identities, AI agents and service accounts, have excessive privileges that create significant risks, far exceeding the 37% of human identities. In addition, nearly half (49%) of identities with administrator-level privileges remain potentially exploitable despite not having been used for 90 days.
The report explains that the combination of excessive privileges and prolonged inactivity among non-human identities has become a particularly apparent security risk in AI and cloud environments. It indicates the importance of properly managing privileges for machine identities, AI agents and service accounts.