Microsoft Warns Chinese-Linked Hackers Used AI Agents to Target Hundreds of Organisations Worldwide

Microsoft has warned that China-linked hackers allegedly used AI-powered tools to automate cyberattacks, exploit trusted cloud applications and gain privileged access to organisations across multiple industries worldwide.

Microsoft Warns of AI-Powered Cyberattacks Targeting Hundreds of Organisations

China-Linked Hacking Group Allegedly Uses AI to Automate Attacks

Artificial intelligence is increasingly being misused for cybercrime, enabling attackers to automate activities that previously required significant manual effort. Microsoft has reported a cyber campaign involving Storm-0940, a hacking group it links to China, which allegedly used an AI-powered tool to target organisations worldwide.

According to the report described in the source material, the attackers automated several stages of their operations, including password attacks, network reconnaissance and attempts to obtain privileged access to cloud environments. In some cases, the campaign reportedly enabled attackers to establish extensive cloud access within hours.

The incident highlights growing concerns about how AI tools can accelerate cyberattacks and make it harder for security teams to detect suspicious activity before significant damage occurs.

ConnectWise Cloud Backup Becomes a Focus of the Investigation

The reported campaign involved ConnectWise, an IT services provider, and its Cloud Backup service. Rather than approaching every organisation independently, the attackers allegedly exploited a trusted software connection to reach customer environments.

ConnectWise CEO Manny Rivelo said a China-aligned group appeared to be responsible for the incident. Following the discovery, the company suspended the affected application connections and began restoring services in stages.

However, disabling a connection does not necessarily remove attackers who may already have gained access to a customer's network. Organisations therefore need to investigate their own systems and determine whether unauthorised access has occurred.

The Chinese Embassy in Washington had not commented on the incident at the time described in the supplied report. The Chinese government has repeatedly denied allegations of involvement in cyber espionage and hacking.

How AI Tools Allegedly Accelerated the Cyberattack

Microsoft's reported findings identify ARTEX, described in the supplied material as an open-source penetration-testing framework, as a tool misused during the campaign.

Penetration-testing tools can help authorised security professionals identify vulnerabilities and evaluate network defences. However, when used maliciously, automation capabilities can also help attackers carry out password attacks, scan systems and exploit weaknesses more rapidly.

According to the report, Storm-0940 used ARTEX's AI capabilities from the early stages of its operations. The tool allegedly automated password spraying and subsequent abuse of OAuth applications, rather than being introduced only after an initial manual intrusion.

Password spraying involves trying a small number of commonly used passwords against multiple accounts. Attackers use this technique to seek access while attempting to avoid triggering account lockout protections.

Trusted Applications and Microsoft 365 Accounts at Risk

The reported campaign also involved the alleged misuse of trusted digital connections between ConnectWise's backup applications and Microsoft 365 accounts.

Because these applications had established permissions, attackers could potentially exploit the trust relationship to access cloud resources without following the usual interactive login process.

Microsoft reportedly observed approximately 2,900 sign-in attempts per minute, with activity reaching more than 2,250 systems. The attackers allegedly used automation to scan networks, alter passwords, establish persistent access and obtain elevated administrator privileges.

These capabilities can make an intrusion particularly difficult to contain. Once attackers obtain administrative access, they may be able to change security settings, access sensitive information or create additional ways to return to a compromised environment.

Although some attempts were reportedly blocked, the campaign allegedly resulted in persistent, high-level cloud access at several organisations.

Multiple Industries Reportedly Targeted Worldwide

The campaign reportedly affected organisations across a broad range of industries, demonstrating that automated cyberattacks can threaten businesses and institutions with very different operational priorities.

The sectors identified in the supplied report include:

• Aerospace and defence

• Automotive and chemicals

• Energy and financial services

• Government and legal services

• Manufacturing and media

• Military and semiconductor industries

• Software and technology

• Think tanks and nonprofit organisations

The breadth of the reported targeting highlights the importance of strong identity management, secure software integrations and continuous monitoring. Organisations that rely on external IT providers may also need to review the permissions granted to third-party applications.

What Organisations Should Do to Protect Their Systems

Microsoft and security specialists have urged organisations connected to the affected service to take precautionary steps and investigate potential compromises.

Disable affected connections

Organisations using ConnectWise Cloud Backup should follow official security guidance and keep affected connections disabled until the provider advises that they are safe to restore.

Review Microsoft 365 security logs

IT teams should investigate unusual sign-in activity, unexpected password changes and newly created administrator accounts or roles. Particular attention should be paid to suspicious OAuth application activity and unusual permission changes.

Preserve evidence during investigations

Security teams should avoid simply deleting suspicious applications. Disabling them while preserving relevant logs and configuration details can help investigators understand how access was obtained and whether attackers remain inside the environment.

Investigate existing access separately

Suspending an application connection does not automatically revoke access already established by an attacker. Organisations should review accounts, credentials, permissions and active sessions, and take appropriate containment measures based on their investigation.

AI Raises New Challenges for Cybersecurity Teams

The reported incident illustrates how AI-assisted automation can increase the speed and scale of malicious activity. Tasks such as password spraying, network scanning and privilege escalation can be coordinated more rapidly when attackers automate multiple stages of an operation.

For defenders, this creates pressure to improve detection systems, strengthen authentication controls and monitor third-party application permissions continuously.

Businesses should also ensure that their incident response plans account for compromises involving trusted service providers. Security teams may need to investigate both the original software connection and any subsequent access established within their own networks.

Conclusion

Microsoft's reported findings highlight the growing cybersecurity risks associated with malicious use of AI tools. The alleged Storm-0940 campaign demonstrates how automated attacks and compromised trust relationships can expose cloud environments across multiple industries. Organisations should prioritise access reviews, security-log monitoring and incident investigations to reduce the risk of persistent unauthorised access.