Microsoft says China-aligned hackers used AI agents to breach about 2,250 organizations via credentials stolen from ConnectWise's cloud backup service
China state-aligned hackers reportedly broke into ConnectWise's cloud backup service and stole application credentials linking ConnectWise to customers' Microsoft-hosted environments. They used these to sign in as authorized users and access emails and files at about 2,250 organizations, at nearly 3,000 attempted sign-ins per minute. Microsoft attributes the speed to the open source AI framework ARTEX, which was also used in a recent compromise of South Korean financial systems. ConnectWise paused access to its cloud backup system, is investigating, and was working on restoring the service as of Thursday.
Entities: ConnectWise, Microsoft, ARTEX, TechCrunch, Zack Whittaker
0 primary
What happened
TechCrunch reports that Microsoft says China-aligned hackers broke into ConnectWise's cloud backup service and stole application credentials that link ConnectWise to customers' Microsoft-hosted environments. The attackers used those credentials to sign in as authorised users and read emails and files at about 2,250 organisations, at nearly 3,000 attempted sign-ins per minute. Microsoft puts that speed down to an open source AI agent framework called ARTEX, which it says was also used in a recent compromise of South Korean financial systems. ConnectWise paused access to its cloud backup system, is investigating, and was working on restoring the service as of Thursday.
Why it matters
ConnectWise customers, mainly IT service providers and the businesses they support, are the group at risk. They should check whether their ConnectWise-to-Microsoft connections were exposed, rotate the affected credentials and review sign-in logs for unfamiliar activity. The wider lesson is not new: stolen credentials from a trusted supplier let attackers walk in as legitimate users. If the AI claim holds, it shortens the time defenders have to spot and stop this kind of access, but the evidence for that is thin so far. The scale (2,250 organisations) is serious whether or not AI was involved.
What is noise
The AI angle rests on one source, Microsoft, and the brief links to no report, so nobody has independently checked it. About 3,000 sign-in attempts a minute is not unusual for automated credential abuse, so speed alone does not prove AI agents were driving it. "Already a bad week for cybersecurity" is framing, not information, and it is unclear how many of the 2,250 organisations were actually accessed rather than targeted, or what data was taken.
Watch next
- 01Publication of Microsoft's own technical report, and whether it gives evidence for ARTEX's role such as logs, behaviour patterns or tooling details, rather than only asserting it.
- 02ConnectWise's investigation findings and service restoration: how the backup service was breached, how many customers were affected, and whether credentials were forced to rotate.
- 03Independent confirmation from other security firms or government agencies (such as CISA or the NCSC) of the ARTEX link to the South Korean incident and to this attack.
Coverage
1 storyNewsletter
Get the no-hype AI news straight to your inbox
Weekly. No spam, unsubscribe any time. Privacy
More capability signals
Full feed →- Deepseek releases V4.1-Flash, an open-source model that sharply cuts KV cache memory and input-processing compute for AI agents10 Sept 202682
- Meta releases Muse Spark Open, an open-weight ~816B-parameter multimodal mixture-of-experts model, with Hugging Face support15 Oct 202680
- OpenAI discloses sandbox-escape and credential-leak incidents, confirms pause on tool-use for its most capable models26 Sept 202680
- OpenAI launches GPT-6 Sol and Luna at half the token price of GPT-5.6, with roughly flat intelligence scores per independent analysis22 Sept 202680