Proofpoint details BluePhishing consent attacks on Entra agent identities
Proofpoint researchers describe BluePhishing, a way to gain first access to a company's Microsoft cloud tenant by tricking users into consenting to a malicious AI agent blueprint. Microsoft Entra administrators should take note.
Written by the SecurityTabs AI newsroom from the original post by Proofpoint Threat Insight; vulnerability scores come from NVD, FIRST and CISA, not the AI.
Proofpoint researchers present a technique they call BluePhishing, which they describe as the next evolution of OAuth consent abuse (tricking a user into approving an app's access to their account). It targets Microsoft Entra Agent Identity, which extends Entra to AI agents. [1]
Proofpoint says Entra Agent Identity closely mirrors the traditional application model, so it inherits many of the same opportunities for abuse. An attacker can set up a blueprint with inheritable permissions and requiredResourceAccess settings, then send a consent link that, once followed, gives a first foothold in the victim's tenant. [1]
According to Proofpoint, consent to a malicious blueprint lets the attacker establish agent identities and operate within the victim tenant. The post is research into the technique; the excerpt available does not report in-the-wild exploitation, a Microsoft fix or a CVE. [1]
Explained
The technology and terms in this story. Learn more
- TechnologyAI agentAn AI model that acts on its own: it plans steps, uses tools and takes actions, not just answers.
- ConceptExploitCode or a technique that turns a software flaw into something an attacker can actually use.
- ConceptCVE · Common Vulnerabilities and ExposuresThe public ID system for vulnerabilities, such as CVE-2026-12345, so everyone means the same bug.
Original sources
- [1]Proofpoint Threat Insight
BluePhishing: Initial Access via Third-Party Agent Blueprint Consent