AI Bots Are Hacking Websites Automatically — And the Fixers Were Crooks
Tuesday, October 20, 2026 · 5-minute read
The iPhone of Vladan Đokić — rector of the University of Belgrade and a leading voice in Serbia's student protest movement — was secretly infected with Pegasus spyware at some point in 2026. The infection used a zero-click exploit, so Đokić didn't have to tap a single thing. The discovery came just six days before Serbia's snap parliamentary election on October 25, raising serious concerns about political surveillance.
↗ The Hacker News / Citizen LabFederal investigators probing the ShinyHunters gang — which hacked the FBI's own jobs website in September — have found that executives at ransomware negotiation firms were apparently working with the hackers, not against them. One executive, Canadian cybersecurity consultant Edward Dubrovsky, was arrested on October 9 and charged with extortion after investigators found he knew about the FBI breach before the FBI did. A second unnamed executive at a large U.S. consulting firm is alleged to have helped run the gang's victim-shaming website.
↗ Krebs on SecurityThe Department of Homeland Security has ordered all federal civilian agencies to stop using DeepSeek AI products and remove them from government systems immediately. The move follows growing concerns in Washington about Chinese-developed AI tools accessing sensitive government data. It's the latest sign that AI software is now firmly in the crosshairs of U.S. national security policy.
↗ BleepingComputerA Chinese-speaking criminal pointed autonomous AI agents at more than 1,200 company websites and let them probe for weaknesses with almost no human involvement. The tool used is called Cairn, an open-source AI hacking assistant. Targets included major names like John Deere, SharkNinja, and Cummins. A security researcher discovered the operation because the attackers accidentally left their control panels exposed on the open internet — a sloppy mistake that let him watch the attacks in real time. It's not yet known how much data, if any, was stolen.
Attackers broke into French cloud giant OVHcloud's CLI tool repository on GitHub and replaced legitimate versions with information-stealing malware. Anyone who downloaded the tool during the window of compromise may have had their developer credentials quietly stolen. This is a supply chain attack — the weapon is the software you already trust.
F5 released emergency out-of-band patches for its widely used NGINX web server. Three of the flaws are rated critical, meaning attackers could run their own code on a vulnerable server — essentially taking it over completely. NGINX is used by a huge slice of the internet, so unpatched servers are a significant target.
Status: Patches available — update immediately. Check F5's official security advisory for affected versions.
The Go programming language team issued an emergency update that blocks eight certificate authority keys from an unnamed provider. Any app built with the new Go version will refuse to connect to websites using certificates from those keys. Mozilla also quietly blocked six related certificates in Firefox around the same time — neither organization has said publicly what went wrong, which suggests something serious.
Status: Go 1.27.3 and 1.26.10 are available now. Developers should upgrade immediately. End users on Firefox are already protected via an automatic background update.
CISA is urging organizations to patch a critical flaw in BeyondTrust Remote Support, a tool IT teams use to access computers remotely to fix problems. A vulnerability in this kind of software is particularly dangerous because it's designed to give deep access to machines — making it a prime target for attackers who want to move quietly through a company's network.
Status: CISA added this to its Known Exploited Vulnerabilities catalog on October 16 — meaning it is being actively used in real attacks. Patch immediately.
Cairn started as a legitimate open-source security research tool — the kind that defenders use to test their own websites for weaknesses before attackers find them. But today's story shows it's now being picked up by criminals who run it autonomously against thousands of targets with almost no effort. Security researcher Snorre Fagerland of NorForge found roughly two dozen malicious Cairn instances by scanning the internet for exposed operator consoles — and could read the attackers' work in real time because they left those panels wide open. This is a stark illustration of how quickly a legitimate research tool can become a mass-market attack weapon when AI removes the skill barrier.
When a company gets hit by ransomware, the first call is often to a specialist negotiator — a professional hired to talk to the hackers, manage the crisis, and ideally lower the ransom. It's a legitimate, growing industry. The ShinyHunters investigation has blown a hole in that idea, revealing that at least two people inside that industry may have been secretly working for the other side the entire time.
ShinyHunters is a hacking group that has extorted more than $70 million from victims this year alone. Their boldest move came in late September when they hacked the FBI's own recruitment website — fbijobs.gov — and threatened to publish the deeply personal personnel files of FBI employees, including medical records, Social Security numbers, and home addresses. The FBI responded fast. Dutch authorities arrested senior suspect Pepijn van der Stap in the Netherlands on September 15, reportedly using flash-bang grenades while he was actively logged into his accounts. Those accounts included a Signal messaging app that was still open — giving investigators a live window into his recent conversations.
What they found in those messages was explosive. Van der Stap had been regularly chatting with Canadian cybersecurity executive Edward Dubrovsky — a man who literally wrote a book recommending that companies always negotiate with hackers. According to investigators, Dubrovsky knew ShinyHunters had stolen data from FBI servers before the FBI itself did, and he never told them. He's now charged with extortion. A second executive at a large, unnamed U.S. consulting firm is alleged to have helped run ShinyHunters' victim-extortion website, shinywire.com, which listed nearly 100 hacked organizations and pressured them to pay up. That executive has not yet been charged.
For everyday people and businesses, the lesson here is uncomfortable: the experts you hire in a crisis may not always have your interests at heart. As ransomware negotiation firms multiply and AI makes attacks cheaper and faster, vetting the people on your own side of the table is no longer optional. Watch for criminal charges against the unnamed U.S. consulting firm executive — that story is likely far from over.