Cyberattacks Without The Old Time Limits: AI’s Growing Role

🔍 Read the full analysis: Cyberattacks Without The Old Time Limits: AI’s Growing Role on ThorstenMeyerAI.com

Before you orderOffer from Amazon

Get the latest gadgets delivered free with Prime

• Fast, free delivery on millions of items
• Prime Video, Amazon Music and more included
• Member-only deals all year

Start your free Prime trial →Free trial for eligible customers · Cancel anytime

As an affiliate, we earn on qualifying purchases.Hide

U.S. agencies and cybersecurity companies have reported documented uses of AI in attacks on industrial controllers and cloud organizations, including scripts aimed at Siemens equipment and an agentic framework used in a campaign linked by Microsoft to Storm-0940. The reports indicate that automation can compress parts of an attack into hours, but they do not show that most attempts succeeded: Microsoft said the majority of the targeting it observed was unsuccessful.

A joint advisory published August 18 by the NSA, CISA, FBI, Department of Energy and Environmental Protection Agency says threat actors used AI-generated exploitation scripts against Siemens S7 programmable logic controllers, or PLCs, in U.S. critical infrastructure. PLCs control industrial processes such as pumps and production lines. The agencies said actors scanned for exposed systems and used scripts disguised as legitimate monitoring tools, built with the open-source snap7 library. They described the activity as an active threat and assessed that it was likely preparation for operational effects.

In an October 21 report, Microsoft Threat Intelligence said Storm-0940, which it describes as aligned with the Chinese state and active since 2021, used the open-source ARTEX multi-agent penetration-testing framework following a compromise of ConnectWise Cloud Backup. Microsoft said the actor used credentials for the backup application to attempt sign-ins to more than 2,250 tenants. It recorded over 151,000 failed logons from October 10 to 14, peaking at 2,900 per minute; none of those attempts succeeded. In several attacks, Microsoft said, the time from initial targeting to privileged cloud access was under three hours.

Microsoft cautioned that most targeting attempts were unsuccessful and that an affected application or successful sign-in does not by itself show that data was accessed. ConnectWise said existing backup archives showed no indication of compromise and that it suspended Cloud Backup, SaaS Security and Cloud Management while investigating. Separately, OpenAI said AI identified critical PHP vulnerabilities that were patched on October 21 before attackers exploited them, and forecast that AI-found critical vulnerabilities would “increase sharply in the coming months.”

Automation Compresses Attack Timelines

The reports point to a practical change in the economics of cyber operations, rather than evidence of an AI system acting as an independently capable “super-hacker.” Scanning, password spraying, adapting scripts and moving through a trusted application are familiar attack steps. AI agents can help carry out or coordinate those steps at a pace that is difficult to match manually. Microsoft said agents can work around the clock; Palo Alto Networks’ Unit 42 described an agent completing hundreds of hours of manual targeting analysis in minutes.

For organizations, that speed matters because a short gap between discovery and privileged access can leave less time to detect unusual activity or contain a compromised account. The risk is particularly consequential when attackers can use a supplier’s trusted access, or when exposed industrial controllers connect to systems that manage physical operations. At the same time, the Microsoft figures show why speed should not be confused with success: its reported password-spray attempts failed, and the company said most of the targeting was unsuccessful.

AI may also increase the volume of flaws defenders need to address. OpenAI’s report describes a defensive benefit in AI finding vulnerabilities before attackers exploit them, but its forecast of a sharp rise in such findings could place more pressure on software makers and customers to verify, prioritize and install fixes quickly. The source material does not establish the scale or timing of any wider increase.

Three Reports, Different Evidence

The three cited developments describe different kinds of evidence. The August government advisory concerns AI-generated scripts and reconnaissance against internet-reachable industrial controllers. Microsoft’s October report describes the use of an agentic framework in a campaign involving a third-party cloud backup service. Unit 42’s July 30 report, meanwhile, describes a Chinese-speaking actor using the DeepSeek model with the Hermes Agent framework, controlled over Telegram. In that account, after an initial exploit failed, the agent searched GitHub for another approach.

These cases do not establish that AI alone caused a breach or that all reported activity resulted in access or damage. They show that AI tools have entered parts of the attacker workflow, while existing weaknesses—exposed equipment, compromised credentials and trusted third-party access—remain central. The same tools can assist defenders: OpenAI says AI identified PHP vulnerabilities that were patched before attackers used them. The central question for organizations is therefore not simply whether to adopt AI, but whether security basics can keep pace with faster discovery and execution.

The agencies’ recommendations for industrial environments focus on controls rather than a new AI product: inventory PLCs, patch firmware, keep controllers off the public internet, separate information technology and operational technology networks, and require multifactor authentication for remote OT access. They also advise monitoring industrial traffic for unexpected write commands.

““The majority of the targeting wasn’t successful.””

— Microsoft Threat Intelligence

Success, Impact and Scale Remain Unclear

The reports do not establish that the ConnectWise-related activity resulted in customer data being accessed or stolen. Microsoft says a successful sign-in or affected app alone does not prove access to data, while ConnectWise says its existing backup archives showed no indication of compromise. The full impact of the investigation and the status of the suspended services are not specified in the source material.

The agencies assess that activity against Siemens PLCs was likely preparation for operational effects; that is an assessment, not confirmation that physical disruption occurred. The source material also does not quantify how many systems were compromised, whether any industrial operation was affected, or how widespread the AI-assisted activity is beyond the cases described. Predictions of a near-term “vulnerability tsunami” are expert opinion rather than a measured trend.

Investigations and Basic Defenses

ConnectWise’s investigation and its decisions about restoring the suspended services remain developments to watch. Organizations that use the affected products should follow updates from the vendor and their security teams, while avoiding assumptions that a service’s suspension confirms customer data theft. The government advisory gives operators a practical checklist: identify every controller, patch firmware, remove direct internet exposure, segment OT networks and secure remote access with multifactor authentication.

For companies more broadly, the recommendations in the source material center on maintaining an accurate asset inventory, setting patch deadlines, reviewing third-party applications and rehearsing incident response. Users can begin with their email accounts by using phishing-resistant sign-in where available and reviewing account access. OpenAI’s forecast also makes vulnerability disclosure and patching timelines worth monitoring, but the scale of any increase is not yet known.

Key Questions

What did U.S. agencies report about AI and industrial systems?

The agencies said attackers used AI-generated exploitation scripts against internet-reachable Siemens S7 programmable logic controllers. They assessed the activity as likely preparation for operational effects, but the source material does not confirm resulting physical disruption.

Did the Microsoft-reported campaign succeed?

Microsoft reported attempts against more than 2,250 tenants and over 151,000 failed logons during a password-spray campaign from October 10 to 14; it said none of those attempts succeeded. The company also said most targeting was unsuccessful and cautioned that a successful sign-in does not necessarily mean data was accessed.

What is ARTEX?

ARTEX is an open-source multi-agent penetration-testing framework. Microsoft said Storm-0940 used it in activity connected to a compromise of ConnectWise Cloud Backup.

Is a surge in AI-discovered vulnerabilities confirmed?

No. OpenAI said AI found critical PHP vulnerabilities that were patched before attackers exploited them, and forecast that AI-found critical vulnerabilities would increase sharply in the coming months. That increase is a forecast, not an established measurement.

What should organizations do now?

Based on the cited advisory and reporting, organizations should inventory exposed systems, patch software and firmware, restrict internet access to industrial controllers, review third-party app permissions, separate IT and OT networks, use multifactor authentication for remote access and practise incident response.

Source: ThorstenMeyerAI.com