Summary: A critical security vulnerability (CVE-2024-38206) in Microsoft’s Copilot Studio allows authenticated attackers to exploit server-side request forgery (SSRF) to leak sensitive information. Microsoft has addressed the issue, but the flaw could potentially impact multiple customers due to shared infrastructure.
Threat Actor: Authenticated attackers | authenticated attackers
Victim: Microsoft | Microsoft
Key Point :
- The vulnerability allows attackers to bypass SSRF protection and access Microsoft’s internal infrastructure.
- Exploitation can lead to the retrieval of managed identity access tokens, granting access to internal resources like Cosmos DB.
- The issue affects multiple customers due to the shared nature of the infrastructure powering Copilot Studio.
- This disclosure follows the announcement of two patched flaws in Microsoft’s Azure Health Bot Service, highlighting ongoing security challenges.
- Microsoft plans to enforce multi-factor authentication (MFA) for all Azure customers starting October 2024 as part of its security initiatives.
Cybersecurity researchers have disclosed a critical security flaw impacting Microsoft’s Copilot Studio that could be exploited to access sensitive information.
Tracked as CVE-2024-38206 (CVSS score: 8.5), the vulnerability has been described as an information disclosure bug stemming from a server-side request forgery (SSRF) attack.
“An authenticated attacker can bypass Server-Side Request Forgery (SSRF) protection in Microsoft Copilot Studio to leak sensitive information over a network,” Microsoft said in an advisory released on August 6, 2024.
The tech giant further said the vulnerability has been addressed and that it requires no customer action.
Tenable security researcher Evan Grant, who is credited with discovering and reporting the shortcoming, said it takes advantage of Copilot’s ability to make external web requests.
“Combined with a useful SSRF protection bypass, we used this flaw to get access to Microsoft’s internal infrastructure for Copilot Studio, including the Instance Metadata Service (IMDS) and internal Cosmos DB instances,” Grant said.
Put differently, the attack technique made it possible to retrieve the instance metadata in a Copilot chat message, using it to obtain managed identity access tokens, which could then be abused to access other internal resources, including gaining read/write access to a Cosmos DB instance.
The cybersecurity company further noted that while the approach does not allow access to cross-tenant information, the infrastructure powering the Copilot Studio service is shared among tenants, potentially affecting multiple customers when having elevated access to Microsoft’s internal infrastructure.
The disclosure comes as Tenable detailed two now-patched security flaws in Microsoft’s Azure Health Bot Service (CVE-2024-38109, CVSS score: 9.1), that, if exploited, could permit a malicious actor to achieve lateral movement within customer environments and access sensitive patient data.
It also follows an announcement from Microsoft that it will require all Microsoft Azure customers to have enabled multi-factor authentication (MFA) on their accounts starting October 2024 as part of its Secure Future Initiative (SFI).
“MFA will be required to sign-in to Azure portal, Microsoft Entra admin center, and Intune admin center. The enforcement will gradually roll out to all tenants worldwide,” Redmond said.
“Beginning in early 2025, gradual enforcement for MFA at sign-in for Azure CLI, Azure PowerShell, Azure mobile app, and Infrastructure as Code (IaC) tools will commence.”
Source: https://thehackernews.com/2024/08/microsoft-patches-critical-copilot.html