🔍 Read the full analysis: The New Cybersecurity Challenge: AI-Driven Attack Speed on ThorstenMeyerAI.com
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
TL;DR
U.S. agencies, Microsoft and Palo Alto Networks have reported cases in which AI tools supported cyber operations, from scripts aimed at industrial controllers to automated targeting and cloud-account attacks. The sources describe faster attack work, but also report failed attempts and do not establish that every affected account or system was compromised. OpenAI’s warning about a sharp rise in AI-discovered vulnerabilities is a forecast, not a measured trend.
A joint advisory published August 18 by the NSA, CISA, FBI, Department of Energy and EPA says attackers used AI-generated scripts in activity targeting internet-reachable Siemens S7 programmable logic controllers, or PLCs. The controllers are used in industrial settings, including manufacturing, energy, water and food production. The agencies said the activity included reconnaissance and capability development, and assessed it as likely preparation to cause operational effects. That is an assessment, not confirmation that physical disruption occurred.
Microsoft Threat Intelligence reported October 21 that Storm-0940, which Microsoft describes as aligned with the Chinese state, used the open-source ARTEX agentic framework after a compromise involving ConnectWise Cloud Backup. Microsoft said the actor attempted logins against more than 2,250 customer tenants and recorded more than 151,000 failed logons in a password-spray campaign from October 10 to 14, peaking at 2,900 failures per minute. Microsoft said none of those password-spray attempts succeeded. In separate cases, it reported that the time from initial targeting to privileged cloud access was under three hours.
Microsoft cautioned that most targeting was unsuccessful and that use of an affected application or a successful sign-in does not necessarily mean data was accessed. ConnectWise, which disclosed the compromise on October 16, said existing backup archives showed no indication of compromise and that it suspended Cloud Backup, SaaS Security and Cloud Management during its investigation. Separately, Palo Alto Networks’ Unit 42 described an AI agent conducting hundreds of hours of manual targeting analysis in minutes. OpenAI said AI had identified critical PHP vulnerabilities patched October 21 and forecast that AI-discovered critical flaws would increase sharply in coming months.
Infographic 01 · The record so far
AI is in the attacker’s toolchain. On the record.
Six documented events in twelve weeks. Each label says what the source itself states about AI. Figures are the vendors’ and agencies’ own.
- 30 Jul 2026Unit 42: AI on attackDeepSeek inside the Hermes Agent framework picks and switches targets on its own. Limited impact: across autonomous and manual attempts, data theft from 3 NetScaler systems confirmed. Single-vendor report.
- 18 Aug 2026NSA, CISA, FBI, DOE, EPA: AI on attackAI-generated exploitation scripts, disguised as monitoring tools, against Siemens S7 PLCs in US critical infrastructure. “Not a theoretical risk.”
- 10–14 OctStorm-0940 password spray: AI on attack151,000+ failed logons across eight organizations. No successful sign-in. Lockouts made up nearly 98% of responses.
- 16–21 OctConnectWise Cloud Backup: AI on attackStolen app credentials, then machine-speed sign-ins to customer tenants. Microsoft attributes it to the China-state aligned Storm-0940.
- 21 Oct 2026OpenAI, Patch the Planet: AI on defenceAI found the critical PHP flaws before attackers did. OpenAI expects similar findings to “increase sharply in the coming months.”
- 21–22 OctQuest Diagnostics: no AI link reportedUnauthorized access to a lab ordering and results system; certain patient records accessed. No cause or attacker named.
Most AI-driven attempts in these reports failed. What changed is the cost of trying.Assessment: Thorsten Meyer
Sources: Unit 42 (30 Jul 2026); NSA, CISA, FBI, DOE, EPA advisory (18 Aug 2026); Microsoft Threat Intelligence (21 Oct 2026); ConnectWise Trust Center (from 16 Oct 2026); OpenAI, Patch the Planet (21 Oct 2026); Quest Diagnostics Form 8-K (22 Oct 2026).
Powered by Thorsten Meyer AI
Faster Operations Raise Response Pressure
The reports point to a practical change in the pace and scale of routine attack steps, rather than evidence of an autonomous system capable of reliably breaking into any target. Reconnaissance, password attempts and adapting scripts can be carried out with less human effort, according to the organizations reporting these cases. That can leave defenders less time to notice activity and contain an intrusion.
The same shift may increase pressure on software teams. If AI tools help researchers identify more vulnerabilities, vendors and customers could face more fixes to evaluate and deploy. OpenAI’s expectation of a sharp increase is a company forecast; the provided source material does not establish a measured future rate. For industrial operators, the stakes also include systems that affect physical processes, making network exposure and separation of operational technology a particular concern.
The reports do not show that AI makes attacks inevitable or that every targeted organization suffered a breach. Microsoft’s figures include a large number of failed attempts. The immediate implication is that organizations should be prepared for faster attempts and should prioritize established controls, rather than treating AI adoption itself as a substitute for security planning.
hardware security key FIDO2 USB-C NFC
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Three Reports Document Different Uses
The reports cover distinct cases and should not be collapsed into a single incident. The August government advisory concerns Siemens industrial controllers and AI-generated scripts. Microsoft’s October account concerns Storm-0940, a third-party cloud application and ARTEX. Unit 42’s July 30 report describes an actor using a DeepSeek model with the Hermes Agent framework and Telegram; the supplied source material does not include the remainder of that case’s details.
Microsoft says Storm-0940 has been active since 2021 and assesses it as China-state aligned. It says the actor used credentials obtained through the ConnectWise Cloud Backup compromise and the trust customers had granted the app. ConnectWise disclosed the compromise October 16; Microsoft published its threat account five days later. These are Microsoft’s findings and attribution, not a public finding from a government investigation in the material provided.
Across the reports, the common factor is automation of familiar tasks, not a claim that AI independently invented a new class of attack. The source material also includes an unnamed security leader’s prediction of a “vulnerability tsunami” within months and a roughly 18-month period to prepare. Those time estimates are expert opinion, not observed measurements or an official forecast.
““This is not a theoretical risk—it is an active threat.””
— Joint advisory from the NSA, CISA, FBI, Department of Energy and EPA
business password manager with multi-factor authentication
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Intrusion Impact Is Not Fully Known
The supplied accounts do not establish that the industrial-controller activity caused physical effects, or specify how many PLCs were successfully accessed. The agencies assess that attackers are likely preparing for operational effects, but that remains an assessment. In the ConnectWise-related activity, Microsoft says some operations reached privileged cloud access, while also warning that a sign-in does not prove data access. ConnectWise said its existing backup archives showed no indication of compromise.
It is also unclear from the supplied material how many organizations experienced confirmed data exposure, what information may have been accessed, or the final scope of ConnectWise’s investigation. The full details of Unit 42’s agent case are not included. Finally, forecasts about a sharp rise in AI-discovered vulnerabilities and a near-term “vulnerability tsunami” have not been verified as outcomes.
small business firewall with intrusion prevention
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Investigations and Security Measures Continue
ConnectWise’s investigation and the status of its suspended services will help clarify the scope and consequences of the cloud backup compromise. Organizations using affected services will need incident updates from their providers and should review access granted to third-party applications. Microsoft’s report gives no confirmed final count of organizations with data accessed.
The joint advisory recommends practical steps for industrial operators: inventory PLCs, apply firmware updates, keep controllers off the public internet, separate IT and operational networks, require multifactor authentication for remote OT access, and monitor for unexpected S7 write activity. It also advises blocking TCP port 102 at the perimeter and using one-way gateways for historian connections where appropriate.
For other organizations, the source material emphasizes asset inventories, patch deadlines, phishing-resistant sign-in, review of third-party applications and rehearsed incident response. Users can begin by securing their email accounts, which can be used to reset access to other services. The next meaningful developments will be confirmed investigation findings, vendor patches and further reporting that distinguishes attempted access from verified compromise.
industrial PLC network security monitoring appliance
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Key Questions
Have AI-driven cyberattacks been confirmed?
Government agencies and cybersecurity companies have reported specific cases in which AI-generated scripts or AI-agent frameworks were used in cyber activity. Those reports do not show that every attempted operation succeeded.
Did the ConnectWise incident expose customer data?
The supplied material does not confirm customer data exposure. Microsoft said some operations reached privileged cloud access but warned that successful sign-in does not necessarily mean data was accessed. ConnectWise said existing backup archives showed no indication of compromise.
What does “under three hours” refer to?
Microsoft said that in several attacks, the time from initial targeting to privileged cloud access was under three hours. It is not a claim that every attack took that long or that all targeted organizations were compromised.
What should industrial operators do now?
The joint advisory recommends inventorying and patching PLCs, keeping them off the public internet, separating operational technology from IT, requiring multifactor authentication for remote access and monitoring for unusual controller traffic.
Is a vulnerability surge already proven?
No. OpenAI forecast that AI-found critical vulnerabilities would increase sharply in coming months. The source material presents that as a forecast, not a measured future outcome.
Source: ThorstenMeyerAI.com
Halloween Picks
halloween
As an affiliate, we earn on qualifying purchases.
