UK school network hit by 3AM ransomware, AI suspected | #ransomware | #cybercrime


A ransomware crew locked roughly 37 computers across a group of UK schools on the morning of September 19, and the security firm Bitdefender now says an AI model probably wrote many of the commands the attacker typed.[1][2] The report went public on November 3, 2026. It matters beyond classrooms: the break-in used ordinary, well known tricks, and the weak spots it exploited exist in many small organisations that run their own computers.
What happened to the school estate
Bitdefender investigated an attack on what it calls a mid-sized UK school estate. It has not named the schools or the trust that runs them, and neither has any news outlet we read.[1][2] The attacker belonged to the 3AM ransomware operation, which surfaced in 2023 and which researchers have linked to the Conti and Royal gangs.[2]
The intruder was active from September 10 until the files were locked on September 19. Bitdefender could not work out how the attacker first got in.[1][2] Two earlier suspicious events, in January and June 2026, may or may not be connected.[1]
How the attack unfolded, according to Bitdefender. Sources: Bitdefender, BleepingComputer.[1][2]
How the attackers got through the network
The first big move was stealing the passwords stored on three backup servers, plus a saved browser login file.[1][2] Those credentials opened more doors. The attacker then created two secret administrator accounts, named library and svc_bkpmon, so they could return even if the first access was closed.[2]
The school estate had grown out of several separate networks that did not trust each other. On September 18 the attacker joined three of them with two-way links, which let a single stolen account reach far more machines.[1][2] Financial files were copied out before the end. On the morning of the 19th the attacker tried to switch off the security software, deleted the machines’ recovery data and ran the encryption program, which locked about 37 of nearly 50 computers reached.[1][2]
Why researchers think an AI model was involved
Bitdefender points to clues in the commands. One call to the backup software used the wrong port and login method, as if written from memory rather than from the manual. Commands from the Linux world turned up inside Windows tools on three separate days.[1] Failed attempts were repeated with small changes, roughly every 40 to 100 seconds.[1][2]
Bitdefender is careful about what this proves. Its report says “What we can document is behavior, and the behavior is that of an agentic loop.”[1] It could not see the attacker’s own machine, so it cannot say which model was used or whether a person was pasting results back and forth.[1][2] It also says nothing in the intrusion was new, and that the attack was slow, with even the quickest retries taking 40 seconds or more.[1]
What the authorities are saying
The report landed the same day as the UK National Cyber Security Centre’s annual review. Its chief executive, Richard Horne, warned of a “real prospect of an unprecedented rise in the number of cyber incidents”.[3] In his words, the cause is “not because AI has invented new ways to attack us”.[3] The agency also expects that, by 2028, criminals will very probably deploy malware driven by AI against Britain’s essential services in a ransomware strike.[3] The school case fits that picture: AI helped with routine work rather than creating a new weapon.
What schools and small organisations should do now
Bitdefender’s advice is practical, and most of it applies to any small business or club that runs its own servers.[1] Treat it as a checklist for your IT supplier.
Defences drawn from the Bitdefender report. Sources: Bitdefender, BleepingComputer.[1][2]
Keep backups out of reach: give them their own accounts, isolate the backup servers and hold at least one copy that cannot be changed or deleted.Ask your IT provider to alert you whenever a new administrator account or a new link between networks appears.Make sure security software runs on every computer, including old ones and any machine nobody manages.Allow remote management tools only from a few known machines.Switch off scripting tools on computers that do not need them.
Ransomware families differ in detail but follow similar steps. Our entries on Akira ransomware, Medusa ransomware and HardBit ransomware show how other groups lock files and what they demand.
What comes next
The schools have not been named, and no source we read says what was stolen beyond the finance files, whether pupils or parents were affected, or whether a ransom was paid. Those answers may emerge if the trust makes a statement or if the gang lists the data on a leak site. The wider lesson is already clear: attackers using AI still rely on old weaknesses, so fixing backups, admin accounts and unmanaged machines remains the best defence.

Click Here For The Original Source.

——————————————————–

............