What To Know About AI’s Effect On Cyberattack Timing

🔍 Read the full analysis: What To Know About AI’s Effect On Cyberattack Timing 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

US agencies and Microsoft have reported separate cases of attackers using AI to automate or assist cyber operations, including scripts aimed at industrial controllers and activity tied to a compromised cloud backup application. The reports show how AI can compress parts of an attack, but Microsoft said most of the targeting it observed was unsuccessful, and the broader effect on attack volume remains uncertain.

In an August 18 joint advisory, the NSA, CISA, FBI, Department of Energy and Environmental Protection Agency said threat actors were using AI-generated exploitation scripts against internet-reachable Siemens S7 programmable logic controllers, or PLCs. These industrial computers can operate equipment such as pumps and valves. The agencies said the scripts were disguised as monitoring tools and built using the open-source snap7 library. They described the activity as an active threat and assessed that it could be preparation for operational effects against critical infrastructure.

Microsoft Threat Intelligence reported on October 21 that the China-state-aligned actor Storm-0940 used the open-source ARTEX agentic framework in activity associated with the compromise of ConnectWise Cloud Backup. Microsoft said the actor used credentials for the backup app to attempt sign-ins to more than 2,250 tenants. A password-spraying campaign from October 10 to 14 generated more than 151,000 failed logons, peaking at 2,900 per minute. In several attacks, Microsoft said, the time from initial targeting to privileged cloud access was under three hours.

Microsoft also stressed that most of the targeting was unsuccessful. A successful sign-in or an affected app did not, by itself, show that data had been accessed. ConnectWise said existing backup archives showed no indication of compromise and that it had suspended Cloud Backup, SaaS Security and Cloud Management while investigating. Separately, OpenAI said AI had identified critical PHP vulnerabilities patched on October 21 before attackers did, and forecast that AI-discovered critical vulnerabilities would increase sharply in the coming months.

Faster Attacks Raise the Patch Burden

The reported cases matter because they show AI being applied to familiar steps in cyberattacks, not because they establish that attackers have developed a new class of autonomous super-hacker. Automating reconnaissance, adapting scripts or attempting many logins can let operators cover more targets or work for longer. Microsoft’s report of privileged access in under three hours in some cases illustrates the limited time defenders may have to detect and contain an intrusion.

AI may also help defenders identify flaws earlier. OpenAI’s account of AI finding vulnerabilities before attackers did points to a possible benefit, but its expectation of a sharp rise in discoveries is a forecast, not a measured trend. If more flaws are found, organizations may face pressure to assess and patch them sooner. The evidence presented here does not establish how often AI will produce successful attacks or how much it changes overall incident rates.

For operators of industrial systems, the stakes include disruption to physical processes as well as data loss. The agencies’ recommended controls focus on established safeguards: know what equipment is connected, keep it off the public internet, separate operational technology from business networks, and protect remote access. For other organizations, the ConnectWise case underscores the risk that a trusted third-party application can provide a route into customer environments.

Three Reports Show Different Uses

The reports describe distinct situations and should not be treated as one coordinated campaign. The August advisory concerns activity against Siemens PLCs in sectors including manufacturing, energy, water, chemicals, food and agriculture, and commercial facilities. The agencies said actors found exposed devices through internet-scanning services and advised organizations to inventory controllers, patch firmware, block unnecessary external access and monitor S7 traffic.

Microsoft’s October account concerns Storm-0940 and the ConnectWise Cloud Backup compromise. ConnectWise disclosed the compromise on October 16, according to the supplied report; Microsoft’s later analysis described how the actor used the app’s credentials and ARTEX. Microsoft said Storm-0940 had been active since 2021. Its report also recorded more than 151,000 failed logons, but those failures are not successful account access and should not be read as evidence that each targeted tenant was breached.

A third case described in the source material comes from Palo Alto Networks’ Unit 42, which reported on July 30 that a Chinese-speaking actor connected the DeepSeek model to the Hermes Agent framework and directed it through Telegram. Unit 42 said the agent searched GitHub after its initial exploit attempt failed. The material supplied here cuts off before the report’s full account, so further details cannot be established from it.

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

— Microsoft Threat Intelligence

Scale and Outcomes Remain Unclear

The reports do not establish how widespread AI-assisted cyber operations are across all attackers, or whether AI was essential to the activity described. Nor do they provide a common measure for comparing attack speed or success with operations conducted without AI. Microsoft’s figures describe specific activity it observed; they do not establish the number of organizations whose data was accessed.

ConnectWise said it found no indication of compromise in existing backup archives, while Microsoft said a successful sign-in alone did not prove data access. The supplied material does not give the final outcome of the company’s investigation. The US agencies assessed that activity against PLCs was likely preparation for operational effects, but the material does not confirm that those effects occurred.

Predictions of a coming “vulnerability tsunami” or a fixed period for organizations to prepare are opinions, not measurements or official forecasts in the cited reports. The timing, scale and net effect of AI on both attacks and defense remain unsettled.

Investigations and Patching Continue

ConnectWise’s investigation and any further Microsoft or company disclosures may clarify whether customer data was accessed and what remediation was completed. Organizations using the affected services should follow their providers’ incident notices and review relevant access logs and application permissions.

For industrial operators, the joint advisory calls for immediate review of controller exposure, firmware, network separation and remote-access protections. For organizations more broadly, practical steps include maintaining an accurate asset inventory, setting patch deadlines, reviewing third-party app access, using phishing-resistant sign-in where available and rehearsing incident response. Users can begin by securing their primary email account, which often supports password resets for other services.

Future reporting and vulnerability disclosures will help show whether AI-assisted operations become more frequent or more successful. For now, the documented cases support attention to existing security controls, while claims about a rapid, sweeping change should be treated as forecasts until comparable evidence is available.

Key Questions

Have attackers used AI in real cyber operations?

US agencies and Microsoft have reported documented cases involving AI-generated scripts against industrial controllers and an AI agent framework used in activity targeting cloud tenants. The reports do not show that every attack step was autonomous.

Did the ConnectWise incident expose customer data?

The material does not confirm customer data theft. ConnectWise said existing backup archives showed no indication of compromise, and Microsoft cautioned that an affected app or successful sign-in does not necessarily mean data was accessed.

What does the under-three-hour figure mean?

Microsoft said that in several attacks, the time from initial targeting to privileged cloud access was under three hours. It is not a reported average for all attacks, and Microsoft said most targeting was unsuccessful.

Should organizations expect a surge in vulnerabilities?

OpenAI forecast that AI-found critical vulnerabilities would increase sharply in the coming months. That is a forecast rather than a confirmed trend; the material provides no measured future count.

What should organizations do now?

Review asset inventories, patch exposed systems, restrict internet access to industrial controllers, separate operational technology from business networks, review third-party app permissions and practise incident response. These steps address risks identified in the reports without relying on predictions about how quickly AI capabilities will change.

Source: ThorstenMeyerAI.com