Companies And Users Face A New Cyberattack Pace With AI

🔍 Read the full analysis: Companies And Users Face A New Cyberattack Pace With AI on ThorstenMeyerAI.com

Buying for a business?Offer from Amazon

Get business pricing on monitors, keyboards and dev gear

• Business-only prices and quantity discounts
• Tax-exempt purchasing
• Multiple users, one account, clear invoices

Create a free Business account →Free to register

As an affiliate, we earn on qualifying purchases.Hide

US agencies and technology companies have reported documented cases of attackers using AI-generated scripts and agentic tools against industrial control systems and cloud services, prompting renewed concern about the pace of cyber operations. The reports describe attacks and attempted access—not proof that AI has made attackers more successful: Microsoft said most of the targeting it tracked failed, and some successful sign-ins did not necessarily mean data was accessed.

A joint advisory published August 18 by the NSA, CISA, FBI, Department of Energy and Environmental Protection Agency says actors used AI-generated scripts against internet-reachable Siemens S7 programmable logic controllers, or PLCs. These industrial computers can control equipment such as pumps and valves. The agencies said the activity involved reconnaissance and capability development, and assessed that it was likely preparation to cause operational effects against critical infrastructure. The advisory named manufacturing, energy, water, chemical, food and agriculture, and commercial facilities as sectors of concern.

Microsoft Threat Intelligence reported October 21 that the China-state-aligned actor Storm-0940 used ARTEX, an open-source multi-agent penetration-testing framework, in activity tied by Microsoft to the ConnectWise Cloud Backup compromise. Microsoft said the actor used credentials associated with the cloud backup app to attempt sign-ins to more than 2,250 tenants. It also recorded more than 151,000 failed logons during a password-spray campaign from October 10 to 14, peaking at 2,900 failures per minute. Microsoft said none of those attempts succeeded.

Microsoft reported that, in several attacks, the time from initial targeting to privileged cloud access was under three hours. It also stressed that most of the targeting was unsuccessful, and that an affected application or successful sign-in does not necessarily establish 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.

AI Compresses Attack Timelines

The reports point to a change in the speed and scale of routine attack tasks, rather than evidence of an autonomous system capable of reliably breaking into any target. Scanning, password spraying, adapting scripts and testing access are familiar techniques. AI tools may let attackers perform more of that work with less manual effort or over longer periods.

That matters to organizations with exposed operational technology or cloud applications that hold broad permissions. A compromised supplier account or trusted app can create a route into many customer environments. When access attempts can be launched quickly and at scale, defenders have less time to detect unusual activity and limit its reach. The Microsoft figures also show why speed should not be confused with success: the recorded password-spray attempts failed, and the company said most targeting did not succeed.

AI is also being used to find software flaws. OpenAI said AI identified critical PHP vulnerabilities that were patched on October 21, before attackers did, and forecast that AI-found critical vulnerabilities would “increase sharply in the coming months.” That is a company forecast, not a measured prediction of how many flaws will emerge or how quickly organizations will be able to patch them.

For companies, the practical pressure is to maintain an accurate asset inventory, apply fixes within defined deadlines, restrict remote access and review permissions granted to vendors and applications. For individuals, protecting email accounts with phishing-resistant sign-in methods can reduce the chance that account access becomes a starting point for wider compromise.

Three Reports, Different Evidence

The industrial-control advisory describes activity that agencies say is ongoing: reconnaissance and AI-assisted script development targeting Siemens S7 PLCs. The agencies warned that exposed controllers are a concern and recommended that organizations inventory devices, patch firmware, keep PLCs off the public internet, separate operational technology from business networks and require multi-factor authentication for remote access. Those measures address the exposure and access paths described in the advisory; they do not establish that every targeted installation was compromised.

The Microsoft report concerns a separate incident involving a trusted cloud application and its credentials. ConnectWise disclosed the Cloud Backup compromise on October 16; Microsoft later described Storm-0940’s use of ARTEX and its attempts to access customer tenants. The report said the framework could carry out several stages of penetration testing with little human interaction, but it did not mean every action was autonomous or successful.

A separate July 30 report from Palo Alto Networks’ Unit 42 described a Chinese-speaking actor using the DeepSeek model with the Hermes Agent framework and directing it through Telegram. The source material says the agent searched GitHub after an initial exploit failed, but does not provide enough detail here to establish the full outcome of that operation. Taken together, these cases document varied uses of AI tools; they do not show that all cyberattacks now rely on AI.

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

— Joint Cybersecurity Advisory from the NSA, CISA, FBI, Department of Energy and EPA, published August 18

Success and Exposure Remain Unclear

The reports do not establish how many organizations suffered confirmed data theft or operational disruption. Microsoft said a successful sign-in or an affected app does not necessarily indicate that data was accessed. ConnectWise reported no indication of compromise in existing backup archives, while saying it had suspended services during its investigation. The available information does not specify the investigation’s final findings or the full extent of any customer impact.

The agency advisory describes targeting and preparation involving industrial controllers, but the source material does not identify confirmed physical effects at facilities. It also does not quantify how many controllers were exposed or compromised. The Unit 42 account, as summarized here, does not give enough detail to assess the outcome of the agent’s attempted exploit.

More broadly, claims about what AI will do next remain forecasts. OpenAI expects AI-discovered critical vulnerabilities to rise sharply, but the report provides no numerical estimate in the supplied material. A security leader’s prediction of a “vulnerability tsunami” within months and a roughly 18-month period of heightened pressure is an expert opinion, not a measured trend or an established timeline.

Investigations and Basic Controls

ConnectWise’s investigation and any subsequent customer guidance will help clarify the scope of the cloud backup incident. Organizations using affected services should follow vendor notices, review account and application activity, and check whether third-party credentials or permissions need to be changed. Microsoft’s report also makes clear that the numbers of attempts alone do not confirm access or data loss.

For critical infrastructure operators, the agencies’ advice centers on practical safeguards: identify PLCs, update firmware, prevent direct internet exposure, segment operational and business networks, require multi-factor authentication for remote access and monitor control-system traffic for unexpected writes. Companies should also review third-party app access and practice incident response so teams can act quickly if suspicious activity appears.

For individual users, the immediate step is to secure the email account that can often reset passwords for other services. Use a unique password and phishing-resistant multi-factor authentication where available, and review recovery details and active sessions. Further reporting is needed to determine how the incidents developed, whether investigations find additional affected customers, and whether AI-assisted activity leads to measurable changes in confirmed intrusion or vulnerability rates.

Key Questions

Did AI-powered attackers successfully compromise more than 2,250 cloud tenants?

No. Microsoft said Storm-0940 attempted sign-ins to more than 2,250 tenants; that number is not a count of successful compromises. Microsoft also said most targeting was unsuccessful and that a successful sign-in does not necessarily mean data was accessed.

What systems did the US agencies warn about?

The August 18 joint advisory described AI-generated scripts targeting Siemens S7 programmable logic controllers used in industrial environments. The agencies warned about controllers reachable from the internet and advised operators to reduce exposure and strengthen remote access controls.

Does the reporting show that AI has made cyberattacks unstoppable?

No. The reports document AI tools used in attack activity, but they do not show that attacks are consistently successful or unstoppable. Microsoft said the majority of its reported targeting failed, and the agencies’ advisory described activity and preparation rather than confirmed physical damage.

What should companies do now?

Organizations should inventory internet-facing systems and industrial controllers, patch known vulnerabilities, restrict remote access, use phishing-resistant multi-factor authentication, review third-party application permissions and rehearse incident response. Operators should also separate operational technology from business networks and monitor for unexpected control-system changes.

What does OpenAI mean by expecting vulnerabilities to rise sharply?

OpenAI said AI found critical PHP flaws patched on October 21 and forecast that AI-found critical vulnerabilities would “increase sharply” in the coming months. That is a forward-looking company statement; the source material does not give a numerical estimate or a confirmed timeline for the increase.

Source: ThorstenMeyerAI.com