Around 37 of nearly 50 machines reached on a UK school estate were encrypted after an affiliate of the 3AM ransomware group executed a campaign that Bitdefender’s researchers say bears the hallmarks of an AI-assisted, automated operation.
Bitdefender’s observations: repeated commands and a suspected agentic loop
Bitdefender analysts flagged a sequence of repeated commands issued at roughly regular intervals — typically every 40 to 100 seconds — where each command was slightly modified from the last. "We were not able to access the machine the attacker worked from, so we cannot document anything about the language model itself: which model it was, whether it ran locally or remotely, or even whether a person was copying and pasting instructions and results between the implant and the model," the company said. The researchers emphasize they could only document the behavior, which looked similar to an agentic loop, and that these observations underpin their conclusion that the threat actor may have used AI-generated commands and an automated agentic loop to carry out the operation.
Credential theft and reconnaissance: modified PowerShell, Veeam, Chrome, Mimikatz
The attackers used a modified public PowerShell tool to extract and decrypt credentials stored on three Veeam backup servers, copied a Chrome database containing credentials, and deployed Mimikatz and LSASS memory-dumping tools. They mapped domain relationships and searched logon-script folders for references to other schools — actions consistent with systematic reconnaissance and credential harvesting across the estate. Bitdefender could not establish how the attacker initially gained access to the environment.
This site is the portfolio.
OSINTSights runs on Cloudflare Workers, D1, R2, and Vectorize, with an AI pipeline on Hetzner ARM. Nubivance designed, built, and operates it. We do the same for clients.
See what we buildPersistence, lateral expansion and final-stage destructive actions
During the intrusion the threat actor created two backdoor domain administrator accounts — named 'library' and 'svc_bkpmon' — and stole financial data from compromised machines. To expand their reach they configured two-way trusts between three previously separate Active Directory domains. In the attack’s final stage they attempted to remove Sophos and Bitdefender protection from reachable hosts, delete system-state backups, and deploy 3AM’s Rust-based encryptor. The encryptor was successfully deployed in approximately 37 cases. Bitdefender traces the malicious activity to a window between September 10 and 19.
3AM ransomware: emergence, links, and a notable past victim
Bitdefender’s write-up places this operation within the public arc of 3AM ransomware, which emerged in 2023 after an operation that deployed a Rust-based encryptor when LockBit was blocked. Researchers later linked 3AM to the Conti and Royal ransomware gangs, and affiliates have been observed using tactics such as email bombing and spoofed IT support calls to gain access to corporate networks. A notable past victim named in the account is Kootenai Health, where an August 2024 attack exposed personal information belonging to more than 464,000 patients.
What this means for school IT teams, technologists, and the NCSC
- School IT teams: Backups and domain boundaries are a primary target — the attackers focused on Veeam credentials, system-state backups, and creating domain trusts to spread. Teams will need to account for credential exposure in backup systems and monitor for unexpected domain trust creation and new domain admin accounts such as the ones named in this case.
- Technologists and security operations: The command pattern Bitdefender documented — repeated, slightly altered commands at 40–100 second intervals — is a behavioral signature to watch for as a possible indicator of automated or AI-assisted activity. Detection efforts should prioritize memory-dump protections, credential safeguard controls, and telemetry that can capture rapid, iterative command sequences.
- The NCSC and policy actors: The National Cyber Security Centre published this year’s "The State of the Cyber Nation" report earlier today and urged that critical infrastructure prepare for AI-powered ransomware attacks by 2028. Bitdefender’s assessment — that affiliates appear to be experimenting heavily with AI tools — aligns with that warning and provides an operational example for preparedness planning.
Bitdefender’s analysis stops short of proving which AI tools, if any, were used; investigators could not access the attacker’s originating machine and therefore could not identify a model or workflow. Still, the pattern of repeated, slightly adjusted commands and the speed of the operation suggest a level of automation the researchers describe as heavy experimentation with AI tools. For UK schools and their suppliers, the record of this intrusion underscores two blunt realities: attackers will target credentials and backups first, and the addition of automation or AI-driven tooling can compress the time between reconnaissance and destructive action. The essential unanswered question the incident leaves is not whether AI can be used, but how rapidly attackers will normalize agentic, iterative tooling in ransomware campaigns before defensive practices catch up.
Source: BleepingComputer — UK school estate hit by 3AM ransomware in suspected AI-assisted attack
