AI’s Impact On Cyberattacks: A Response Plan For Companies And Users

🔍 Read the full analysis: AI’s Impact On Cyberattacks: A Response Plan For Companies And Users on ThorstenMeyerAI.com

Before you orderOffer from Amazon

Get everyday essentials 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

US agencies and cybersecurity researchers have documented attackers using AI-generated scripts against industrial controllers and AI agents to accelerate attempts against cloud accounts, adding urgency to calls for companies to strengthen basic defenses. The reports describe faster execution of familiar attack steps, not evidence that AI has made attacks automatically successful: Microsoft said most targeting in one campaign failed.

A joint advisory published Aug. 18 by the NSA, CISA, FBI, Department of Energy and EPA says threat actors used AI-generated exploitation scripts against internet-reachable Siemens S7 programmable logic controllers, or PLCs. PLCs operate industrial equipment such as pumps, valves and production lines. The agencies said the activity involved reconnaissance and capability development and assessed that it was likely preparation for operational effects against critical infrastructure. That assessment describes the agencies’ view of the activity; it does not establish that the targeted facilities suffered operational damage.

Microsoft Threat Intelligence reported Oct. 21 that Storm-0940, which Microsoft describes as a China-state-aligned actor, used the open-source ARTEX agentic penetration-testing framework in activity involving the ConnectWise Cloud Backup compromise. Microsoft said the actor used credentials for the cloud backup app to attempt sign-ins to more than 2,250 tenants. Between Oct. 10 and 14, Microsoft recorded more than 151,000 failed logons in a password-spraying campaign, peaking at 2,900 per minute. Microsoft said none of those attempts succeeded and that, in several attacks, the time from initial targeting to privileged cloud access was under three hours.

Microsoft cautioned that most of the targeting was unsuccessful and that an affected app or successful sign-in does not by itself prove data was accessed. ConnectWise said existing backup archives showed no indication of compromise and said it suspended Cloud Backup, SaaS Security and Cloud Management during its investigation. Separately, Palo Alto Networks’ Unit 42 reported that an AI agent performed hundreds of hours of manual targeting analysis in minutes. OpenAI said AI had identified critical PHP vulnerabilities patched Oct. 21 before attackers did, and predicted such findings would increase sharply in coming months.

Faster Attacks Raise the Cost of Delay

The reports matter because they describe AI reducing the time and effort needed for tasks that already feature in cyberattacks: scanning systems, trying passwords, adapting code and investigating targets. A faster attack can leave defenders less time to detect activity and respond, particularly where one trusted application has access to many customer environments. Microsoft’s account also shows why speed should not be confused with success: it reported a large volume of failed logons alongside a smaller number of incidents reaching privileged access.

For industrial operators, the concern extends beyond stolen data. A compromised controller could affect physical processes, though the advisory describes assessed preparation and active targeting, not confirmed operational disruption. For ordinary users, email and other accounts remain important entry points: attackers can use stolen credentials or convincing messages without needing a novel AI breakthrough. The practical response is to shorten exposure through patching, strong authentication, restricted access and rehearsed incident procedures.

Three Reports, Familiar Attack Steps

The Aug. 18 advisory names Siemens S7 models from the S7-200 through the S7-1500, including safety controllers. The agencies said actors found exposed PLCs using internet-scanning services and used scripts built with the open-source snap7 library, disguised as legitimate monitoring tools. They described AI as reducing the technical expertise and time required for this activity. Their recommended measures include inventorying controllers, patching firmware, keeping PLCs off the public internet, blocking TCP port 102 at the perimeter and separating operational technology from corporate IT.

Microsoft’s Oct. 21 account followed ConnectWise’s Oct. 16 disclosure of a Cloud Backup compromise. Microsoft said ARTEX can support activity from reconnaissance to data exfiltration with little human interaction. The company also stressed that the campaign’s broad targeting mostly did not succeed and that access to an app does not necessarily mean customer data was reached.

Unit 42’s July 30 report provides another example of AI-assisted activity: it described a Chinese-speaking actor connecting the DeepSeek model to the Hermes Agent framework and directing it through Telegram. These cases do not establish that every AI tool or model is being used maliciously. They do show that existing security weaknesses can be exploited with greater automation.

“This is not a theoretical risk—it is an active threat.”

— Joint advisory by the NSA, CISA, FBI, Department of Energy and EPA

What the Reports Do Not Establish

The public accounts do not establish that the industrial-controller activity caused physical disruption, or how many facilities were affected. Microsoft said its reported targeting mostly failed and that an affected app or successful login does not necessarily mean data was accessed. ConnectWise said its existing backup archives showed no indication of compromise, but the companies’ investigations and the full scope of any exposure remain relevant to understanding the incident.

Claims about a coming “vulnerability tsunami” are forecasts, not measured facts. A security leader cited in the source material predicted conditions could worsen before improving and estimated roughly 18 months of pressure, but that is an expert opinion without a stated measurement method. OpenAI’s expectation that AI-found vulnerabilities will rise sharply is also a forecast. The sources do not quantify how much AI changes the overall rate of successful cyberattacks.

Immediate Steps for Defenders

For industrial operators, the joint advisory recommends identifying every PLC, applying firmware updates, removing direct internet exposure, blocking TCP port 102 where appropriate, separating IT and operational networks, requiring multifactor authentication for remote OT access and monitoring for S7 write activity outside approved change windows. Companies should review third-party applications and the permissions they hold, especially where a trusted service connects to multiple customer tenants.

Across organizations, security teams should set patch deadlines, use phishing-resistant sign-in where available, review privileged accounts and rehearse incident response. Users can begin with their email accounts: use a unique password stored in a password manager, turn on multifactor authentication, check recovery options and treat unexpected sign-in prompts or links cautiously. These measures do not remove risk, but they reduce common routes into accounts and systems.

Organizations should track updates from the agencies, Microsoft and ConnectWise as investigations and advisories develop. The next useful evidence will be confirmed findings about any data access or operational impact, along with details on fixes and indicators defenders can use. Until then, the reports support action on established security basics rather than treating forecasts as settled outcomes.

Key Questions

Have AI-assisted attacks caused confirmed damage to industrial equipment?

The joint US advisory describes targeting and preparation involving Siemens controllers, but the supplied reporting does not confirm physical damage or operational disruption.

Did the Microsoft-reported campaign access all 2,250 tenants?

No. Microsoft said the actor attempted sign-ins to more than 2,250 tenants and that most targeting was unsuccessful. It also said an affected app or successful sign-in does not necessarily mean data was accessed.

What should companies do first?

Start with an accurate inventory, timely patching, phishing-resistant multifactor authentication, restricted third-party app permissions and tested incident-response procedures. Industrial operators should also keep controllers off the public internet and separate operational technology from corporate IT.

What can individual users do?

Secure email first with a unique password and multifactor authentication, review account recovery settings and be cautious with unexpected links or sign-in prompts.

Is a “vulnerability tsunami” confirmed?

No. That phrase and the estimate of roughly 18 months of heightened pressure are an expert forecast cited in the source material. They are not a measured finding or a confirmed timeline.

Source: ThorstenMeyerAI.com