Power Automate Desktop CVE-2026-40374

Exploitation requires a valid authenticated session with at least low-privilege access (PR:L per CVSS vector AV:N/AC:L/PR:L/UI:N). The attacker must be able to reach the affected Power Automate for Desktop instance over a network (AV:N). No special non-default configuration is indicated as a prerequisite based on available data. Key limiting factors: the attacker must already possess valid credentials for the target environment, the attack is classified as not automatable by SSVC (ruling out mass exploitation), and the scope is unchanged (S:U), meaning the impact is contained to the vulnerable component itself. Organizations where Power Automate for Desktop is isolated to workstations without network-accessible interfaces, or where it does not handle highly sensitive data, face reduced practical risk.

The CVSS vector AV:N/AC:L/PR:L/UI:N/S:U/C:H/I:N/A:N describes a low-complexity, network-accessible flaw requiring only low-privilege authentication, with high confidentiality impact - metrics that could suggest elevated priority. However, multiple signals substantially reduce practical urgency: EPSS is only 0.11% (29th percentile), indicating very low observed exploitation probability relative to the broader vulnerability population. CISA KEV status is absent. SSVC assessment confirms exploitation status as 'none', classifies the flaw as not automatable, and characterizes technical impact as only partial. Together these signals indicate this is a real but lower-priority vulnerability - important to patch in environments where Power Automate for Desktop handles sensitive data, but unlikely to be targeted opportunistically. No CVSS signal conflicts are observed; the PR:L requirement meaningfully limits the attacker pool to those already authenticated.

An attacker with low-privilege credentials on the target network - such as a standard domain user in an enterprise environment - queries or interacts with Power Automate for Desktop over the network in a manner that causes the application to return sensitive data beyond what that user is authorized to access. No publicly available exploit code or proof-of-concept has been identified at the time of analysis, and SSVC indicates the attack is not automatable, suggesting exploitation requires some manual interaction or knowledge of the application's behavior.