What the FBI warned: In a November 13 advisory, the bureau said attackers are using a tool called ARTEX to automate stages of cyber intrusions. In some cases, intruders reached privileged access in under three hours.
Are the outages connected? We found no evidence in the advisory or the incident reports reviewed that ARTEX caused the recent Apple, Microsoft 365 or Chase disruptions. Those organizations reported technical issues. The investigations and public explanations have different levels of detail.
The FBI Warned About ARTEX AI Attacks. Did They Cause the Recent Outages?
A real FBI warning about AI-assisted cyberattacks is circulating alongside claims about the Apple, Microsoft and banking disruptions. The warning describes serious attacks, but the available evidence does not establish that ARTEX caused those outages.

Jump to a section
First came the outages. Microsoft services faltered. Bank customers reported delayed transactions. Apple users struggled to sign in. Then a screenshot of an FBI warning about an AI-powered hacking tool began appearing alongside the disruptions.
It is an alarming combination, and it deserves a careful answer. The FBI warning is genuine. The leap from that warning to a coordinated attack on the services that recently went down is not supported by the evidence we found.
AI-assisted attacks are real. So are software failures. When a warning and an outage occur in the same week, the timing alone does not establish a connection. We checked the sources on November 17, 2026.
What the FBI actually warned about
The FBI's November 13 FLASH advisory describes ARTEX as a framework that uses AI agents to automate parts of security testing. A penetration test is an authorized effort to find weaknesses in a system. Tools built for that purpose can also be abused by attackers.
According to the FBI, malicious actors have used ARTEX since at least August against organizations in sectors including finance, technology, healthcare and government. The bureau says some intrusions moved from initial targeting to privileged access in less than three hours. Privileged access means permissions powerful enough to administer or make significant changes to systems.
“Some” matters. The warning does not say every attack takes three hours, that ARTEX can break into any system, or that a machine independently chose to attack the companies experiencing outages. The framework connects to external AI models and can carry out tasks set by an operator with limited human intervention. The FBI uses the term “super intelligence” for its agents, but the advisory does not establish that the system possesses artificial superintelligence.
The screenshot shared with Hacked.com describes a real warning and refers to a real ConnectWise incident. We cannot verify the authenticity or origin of the social-media post from the screenshot alone. Its proximity to outage claims is not evidence that the incidents share a cause.
The Microsoft connection is real, but it is a different incident
The detail most likely to cause confusion involves Microsoft 365. In October, a threat actor used compromised application credentials associated with ConnectWise Cloud Backup, according to Microsoft's security research. ConnectWise provides backup and other services to organizations that may have authorized its applications to access their Microsoft 365 environments.
Think of an application credential as a key issued to a piece of software rather than a person. If an attacker obtains that key, they may be able to act through an existing trusted connection. A password reset for one employee does not necessarily revoke the application's access. That is why the list of trusted applications matters as much as the list of human users.
Microsoft attributes the campaign to a China-state-aligned group it calls Storm-0940 and says ARTEX accelerated its activity. The company observed unauthorized authentication and follow-on actions in multiple customer environments. It also said the activity did not result from a compromise of Microsoft corporate systems, products or services. ConnectWise's advisory says it identified unauthorized use of application credentials and is investigating how those credentials were obtained.
What does the 2,250 figure mean?
The FBI advisory and the Microsoft research it cites both say the attacker attempted to sign in to more than 2,250 downstream tenants after the ConnectWise compromise. A tenant is an organization's separate Microsoft 365 environment.
An attempted sign-in is not proof that every tenant was breached or that data was stolen. Microsoft reports successful authentication and follow-on activity in multiple customer environments. It also says that even a successful authentication does not, by itself, establish that customer data was accessed or stolen.
The distinction cuts both ways: it would be wrong to call all 2,250 organizations confirmed victims, and it would be equally wrong to dismiss the incident as merely failed attempts. Some customer environments did see further suspicious activity and require individual investigation.
Did ARTEX cause the Apple, Microsoft or banking outages?
Based on the sources available when we checked, there is no evidence connecting ARTEX to those recent outages. The FBI warning discusses attacks and the earlier ConnectWise compromise; it does not attribute the subsequent service disruptions to ARTEX. Here is what the affected organizations have said:
- Microsoft 365: Microsoft attributed the November disruption to a technical issue involving a network upgrade and said it had seen no indication of a cyberattack or activity by AI agents, AP reported. The October ConnectWise breach and the November Microsoft 365 outage are separate events.
- Chase and bank payments: Chase said a software bug affected its systems and that the issue was not cyber-related, according to AP. Our banking outage report explains the effect on some transactions involving other banks.
- Apple: Apple told AP that an internal update caused service disruptions. MacRumors reported that Apple told Bloomberg the incident was not a cyberattack. Read our Apple outage and account-safety guidance.
- NSA Net: The NSA said internal systems experienced temporary degradation during scheduled maintenance. Two officials told CBS News there was no evidence of an attack at the time. Our NSA Net report distinguishes the agency's statement from what remains unconfirmed.
These are attributed explanations, not independent forensic reports by Hacked.com. New evidence could change the picture. But a shared week on the calendar, a viral post and an earlier breach do not establish a shared attacker, an AI cause or a coordinated campaign.
The other connection: stolen keys can pay for the next attack
There is a genuine connection to other AI-security research, but it is not a connection to the outages. On November 13, Amazon Threat Intelligence described records from AI agents interacting with its decoy systems. One set of records described stolen AI service keys being used to fund further searches for credentials.
Amazon assessed with moderate confidence that the campaigns used ARTEX. It also cautioned that it could not independently verify outcomes the agents described on systems outside its sensors. Our report on stolen AI keys and credential hunting explains that distinction. Amazon's observations and the ConnectWise incident are separate investigations; using the same framework does not establish they were run by the same operator.
Put the two findings together and the practical concern becomes clearer. A compromised service can provide computing power or credentials; a trusted integration can provide access. Automation can then reduce the time a security team has to notice and interrupt misuse. That is an analysis of the combined research, not proof that ARTEX caused a particular outage.
What organizations should do
Start with the access your organization actually uses. An ARTEX warning is not a reason to delete every connected application or shut down every service.
- Inventory trusted applications and their permissions. Ask your Microsoft 365 administrator or managed service provider which applications can access each environment, including old integrations that may still be authorized. An application identity can operate without an employee approving a fresh sign-in.
- If you used ConnectWise Cloud Backup, follow current vendor guidance. Have the responsible administrator review affected tenants and historical application activity, including environments where the integration was used previously. ConnectWise says vendor-side containment does not remove any unauthorized access already established in a customer environment.
- Investigate before deleting evidence. Review application sign-ins, credential additions, permissions and privileged-role changes alongside legitimate administrative activity. Microsoft's guidance says not to delete enterprise applications as the first response. Preserve relevant logs and configuration where possible, but do not delay urgent containment.
- Remove confirmed unauthorized access and rotate affected credentials. Coordinate with your incident-response team and provider. Check for newly created credentials or permissions that could survive remediation of the original application. Re-enable a service only under the vendor's current restoration guidance.
- Protect the ordinary doors too. Keep software updated, restrict high-privilege access, use multifactor authentication for people and monitor account and application activity. MFA for employees is valuable but does not automatically prevent misuse of a stolen application credential.
For more detail on mailbox and application checks, see our Microsoft 365 security guide. Small teams can use our business security baseline to prioritize ownership, access reviews and tested recovery.
What individuals should do
If an app or bank service is unavailable, check its official status page or contact it using a number you already know. An outage alone is not proof that your account was accessed. Avoid unexpected messages claiming you must urgently verify your password or pay to restore access.
If you notice an unauthorized sign-in, unfamiliar account changes or a payment you did not make, treat that as a separate security event. Contact the provider or bank through an established channel, secure the account from a trusted device and review our account recovery guide. Your next step should match what actually happened, not the most alarming explanation online.
The FBI's warning is worth taking seriously because automated attacks can move quickly through trusted connections. The recent outages deserve scrutiny too. Treating them as the same incident without evidence obscures the actual lesson in both: know which systems you depend on, who can access them and how you will recover when something goes wrong.