Companies, Users, And The Faster Cyberattack Threat From AI

🔍 Read the full analysis: Companies, Users, And The Faster Cyberattack Threat From AI on ThorstenMeyerAI.com

Before you orderOffer from Amazon

Get audio and creator gear 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 cyber operations in which attackers used AI-generated scripts or agent frameworks to target industrial controllers and cloud organizations. The reports show that software can speed up reconnaissance and credential attacks, but they also say many attempts failed and do not establish that every targeted system or account was compromised.

A joint advisory published August 18 by the NSA, CISA, FBI, Department of Energy and Environmental Protection Agency says attackers have used AI-generated exploitation scripts against internet-reachable Siemens S7 programmable logic controllers, or PLCs. PLCs are industrial computers used to control equipment and processes. The advisory names models from the S7-200 through the S7-1500, including safety controllers, and says the activity has targeted sectors such as energy, manufacturing, water and food and agriculture.

According to the agencies, attackers identify exposed controllers through internet scanning services and use scripts built on the open-source snap7 library, disguised as legitimate monitoring tools. The agencies assess that the activity may be preparation for operational effects, rather than reporting that such effects have occurred. Their advisory recommends keeping controllers off the public internet, patching firmware, separating operational technology from business networks and monitoring industrial traffic.

In an October 21 report, Microsoft Threat Intelligence said the China-state-aligned actor Storm-0940 used ARTEX, an open-source multi-agent penetration-testing framework, in activity linked by Microsoft to the ConnectWise Cloud Backup compromise disclosed by ConnectWise on October 16. Microsoft says the actor attempted sign-ins to more than 2,250 tenants using credentials associated with the cloud backup app. In a password-spraying campaign from October 10 to 14, Microsoft recorded more than 151,000 failed logons, peaking at 2,900 per minute; none of those attempts succeeded. In several attacks, Microsoft says, 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 necessarily mean data was accessed. ConnectWise said existing backup archives showed no indication of compromise and said it suspended Cloud Backup, SaaS Security and Cloud Management while investigating. Separately, cybersecurity firm Palo Alto Networks’ Unit 42 described an AI agent that completed extensive targeting analysis in minutes. These reports describe particular cases; they do not establish that AI has made all attacks more effective.

Faster Workflows Put Defenders Under Pressure

The reported change is less about a new class of attacker than about the time and effort required for routine attack steps. Scanning exposed systems, trying credentials and adapting scripts can be repeated by automated tools for longer periods and across more targets. Microsoft says the framework it observed could support activity from reconnaissance through data exfiltration with limited human interaction, although the report does not mean every stage succeeded in these incidents.

For organizations, speed can narrow the time available to detect unusual access and respond. A cloud app already trusted by customers can also become a route into multiple tenants if its credentials are compromised. In industrial environments, a reachable controller creates a different risk: activity against operational technology may affect physical processes. The agencies’ warning about possible operational effects is an assessment, not confirmation that production or public services were disrupted.

AI can also assist defenders. OpenAI said it used AI to identify critical PHP vulnerabilities that were patched on October 21, before attackers found them, and forecast that AI-discovered vulnerabilities would rise sharply in coming months. That is the company’s expectation, not a count of future flaws. More discoveries could mean faster fixes, but also more work for organizations responsible for testing and applying patches.

The immediate response does not depend on buying an AI product. The source reports point to asset inventories, timely patching, phishing-resistant sign-in, third-party app reviews and practiced incident response. For industrial operators, network isolation and restricted remote access are also central. For individual users, securing the email account is a useful starting point because it often controls password resets and access to other services.

Three Reports, Different Attack Paths

The reports cover distinct events and should not be treated as evidence of one coordinated campaign. The August government advisory concerns activity against Siemens industrial controllers. Microsoft’s October report concerns Storm-0940, the ARTEX framework and the ConnectWise Cloud Backup compromise. Unit 42’s July 30 report describes a separate actor using a model with the Hermes Agent framework and directing it through Telegram. The available material does not establish a connection among these cases.

In Microsoft’s account, the failed password-spray attempts are separate from the successful access described in several other attacks. The failed logons peaked at 2,900 per minute, but Microsoft says none succeeded; the under-three-hour figure applies to several attacks and describes the reported time to privileged cloud access. Keeping those details separate matters: a large volume of attempts is not proof of account access, and access is not proof that information was taken.

The source material also cites a security leader’s forecast of a coming “vulnerability tsunami” and a period of roughly 18 months to prepare. That is an individual’s opinion, not a measured trend or an official timetable. The documented cases support concern about faster automated activity, but do not by themselves quantify how widely AI is being used or how much it changes overall breach rates.

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

— Microsoft Threat Intelligence

How Much Activity Succeeded

The public accounts leave important questions open. Microsoft has not established in the cited material whether data was accessed in every incident involving the affected app or whether the ConnectWise investigation found evidence beyond the statement about existing backup archives. The scope and consequences of the reported controller activity are also unclear; the agencies describe a threat and possible preparation, not confirmed physical disruption.

It is also not clear how representative these cases are of cyberattacks overall. The reports do not provide a common measure of how often attackers use AI, whether AI materially raises success rates, or how much time it saves across different kinds of operations. Forecasts of a sharp increase in vulnerabilities or a near-term surge in attacks should be treated as projections, not established outcomes.

Investigations and Patching Continue

ConnectWise said it suspended the named services while investigating; the next material update would clarify the investigation’s findings, affected customers and whether any data was accessed. Organizations using the services should follow vendor notices and review their own access and incident records rather than infer compromise from a failed attempt or app exposure alone.

Operators of Siemens S7 systems can act on the agencies’ published guidance now: inventory controllers, patch firmware, block unnecessary internet access, restrict remote connections and monitor for unexpected writes. Companies should review third-party app permissions and strengthen account protections. As AI-assisted vulnerability research develops, security teams will need to track vendor advisories and prioritize fixes based on exposure and risk. The evidence so far supports practical security work, while the scale of the broader change remains to be established.

Key Questions

Have attackers used AI against industrial systems?

US agencies say they have, citing AI-generated scripts used against internet-reachable Siemens S7 controllers. The advisory warns of active activity but does not report confirmed physical disruption.

Did the ConnectWise-related activity expose customer data?

The cited information does not establish that customer data was accessed in every incident. Microsoft says a successful sign-in does not necessarily indicate data access, and ConnectWise said existing backup archives showed no indication of compromise while its investigation continued.

What does the under-three-hour figure mean?

Microsoft says that in several attacks, the time from initial targeting to privileged cloud access was under three hours. It is not a claim that every attempt succeeded in that time.

What should companies do now?

Start with known safeguards: inventory systems, patch exposed software, require strong authentication, review third-party app access and practice incident response. Industrial operators should also keep PLCs off the public internet and separate operational technology from business networks.

Is a surge in AI-discovered vulnerabilities confirmed?

No. OpenAI said it expects AI-found critical vulnerabilities to increase sharply, but that is a forecast. The available reports do not quantify the future number of vulnerabilities or establish a general increase in attack success.

Source: ThorstenMeyerAI.com