Vulnerability in Microsoft's Copilot might have facilitated undetectable hacking attack without user intervention

Hackers might exploit the identified flaw, named "EchoLeak," to obtain user data without requiring any direct user engagement.

, and Administrator

2025 September 13 . 12:51 PM

2 min read

Microsoft's Copilot faces a potential security breach, enabling a zero-click assault without user... — Microsoft's Copilot faces a potential security breach, enabling a zero-click assault without user interaction

Vulnerability in Microsoft's Copilot might have facilitated undetectable hacking attack without user intervention

In a recent development, tech giant Microsoft has taken steps to mitigate the EchoLeak issue, a critical vulnerability discovered in its Copilot AI tool. This vulnerability, identified as CVE-2025-32711, posed a significant risk to organisations worldwide due to the default configuration of Microsoft Copilot.

The EchoLeak vulnerability could have allowed hackers to infiltrate the system and steal sensitive data from Microsoft 365 Copilot without any user interaction. This stealthy attack method is unprecedented, marking EchoLeak as the first known zero-click attack on an AI agent.

Microsoft released an advisory on Wednesday, assuring customers that the issue is fully addressed, and no further action is necessary. The organisation that discovered and reported the security vulnerability is Check Point Software Technologies, with researchers at Aim Security being credited for their findings.

Jeff Pollard, vice president and principal analyst at Forrester, believes the EchoLeak vulnerability aligns with prior concerns about the potential security risks from AI agents. According to Pollard, once an AI agent is empowered to operate on behalf of a user, attackers may find ways to exploit it due to the wealth of information in work and personal email accounts.

The attack could exploit an "LLM scope violation," enabling untrusted input from outside an organisation to commandeer an AI model and access privileged data. This includes chat histories, OneDrive documents, Sharepoint content, Teams conversations, and preloaded data from an organisation.

In response to this security incident, Microsoft is implementing defense-in-depth measures to further enhance its security posture. The company appreciates Aim Labs for identifying and responsibly reporting the EchoLeak vulnerability.

As the world continues to embrace AI technologies, it is essential for organisations to remain vigilant and proactive in addressing security concerns. The EchoLeak incident serves as a reminder of the importance of robust security measures in the AI era.