$388M Crypto Heist, AI-Powered Hackers, and a "Reformed" Criminal's Double Life
Monday, September 28, 2026 · 5-minute read
Dutch authorities arrested 23-year-old Pepijn van der Stap on or around September 16, suspecting him of helping the hacker group ShinyHunters steal data and extort victims. Van der Stap had previously been convicted of similar crimes in 2023, served time, and was working in cybersecurity when he was picked up again. Within days of his arrest, the remaining ShinyHunters members escalated dramatically — stealing data from the FBI and extorting the Russian ransomware group Cl0p.
↗ Krebs on SecurityCameron Wagenius, a 22-year-old U.S. Army soldier stationed in South Korea, was sentenced to nearly six years in prison today. Using the online handle "Kiberphant0m," he stole call and text metadata for over 100 million AT&T customers and extorted at least ten tech and telecom companies. He was also ordered to pay almost $300,000 in restitution to his victims.
↗ Krebs on SecurityShinyHunters is actively exploiting a flaw in Oracle PeopleSoft — and they found a clever way around the safety nets meant to block it. They use a URL-encoding trick to slip past web application firewall rules, letting them continue attacking servers that haven't been patched. If your organization runs Oracle PeopleSoft, this is urgent.
↗ BleepingComputerOn September 24, an attacker drained roughly $388 million from cryptocurrency exchange Bitget. The entry point wasn't Bitget itself — the attacker found a vulnerability in a third-party security product Bitget relied on. That flaw handed the attacker high-level internal credentials, which they then used to send fake withdrawal commands to Bitget's hot wallets. Cold wallets — the offline storage holding most customer funds — were not touched. Bitget's CEO confirmed the details in a livestream today and said the exchange has resumed Bitcoin withdrawals. Suspected North Korean hackers have been linked to the theft.
A group called JADEPUFFER is running a new kind of ransomware playbook — one that uses AI agents to do the dirty work. They break into Microsoft Azure environments, scout around automatically, steal credentials, and then delete core cloud infrastructure. Instead of just encrypting files for ransom, they destroy the environment entirely, leaving victims scrambling to rebuild. Security researchers found they gained access by compromising service principals — the automated accounts that applications use to operate inside cloud platforms.
This flaw lets an attacker who isn't logged in at all send malicious commands that the server will execute. Citrix NetScaler ADC and Gateway are widely used by companies to manage secure remote access — meaning this flaw sits right at the front door of many corporate networks. Attackers are already exploiting it in the wild. CISA has ordered U.S. government agencies to patch by Wednesday.
Status: Patch available — update immediately.
A companion flaw to CVE-2026-88771, this one allows remote code execution or can knock the system completely offline with a denial-of-service attack. Both vulnerabilities are being actively used by attackers right now. The same Citrix patch covers both — if you or your organization uses NetScaler, applying this update is the top priority today.
Status: Patch available — apply immediately alongside CVE-2026-88771 fix.
This vulnerability in Oracle PeopleSoft — software used by universities, government agencies, and large companies to manage HR and finance — is being actively exploited by ShinyHunters. Even organizations that deployed a WAF to block attacks are at risk, because the hackers found a way to encode their requests to sneak past those filters. Patching the underlying software is the only reliable fix.
Status: Patch available — WAF rules alone are not sufficient protection.
Security researchers at Cleafy uncovered a disturbing new feature in a malware-as-a-service operation called RatHat. The Android banking trojan installs itself on victims' phones, overlays fake login screens on banking apps, and harvests passwords and text messages. The new twist: the criminals running it feed that stolen data into Google's Gemini AI model to estimate each victim's bank balance. Gemini then sorts victims into "high-value" and "mid-value" buckets so the criminals know whose accounts are worth targeting first. The AI isn't moving money itself — it's acting as a triage tool to help criminals work more efficiently. Cleafy has tracked nearly 100 deployments of the RatHat control panel since April 2026, suggesting this is spreading fast.
Researchers at ThreatDown discovered a new botnet called Carbonato that breaks into poorly secured server infrastructure and installs an AI agent framework called Hermes Agent. Once installed, the agent listens for commands sent through Telegram — a regular messaging app — and carries out tasks like stealing credentials and spreading to neighboring servers. The attack targets Docker daemons left exposed without a password on the open internet. It scans for new victims every five minutes, so it spreads quickly. This is an early example of criminals weaponizing legitimate AI agent frameworks for infrastructure attacks.
The story of Pepijn van der Stap reads like something from a thriller novel. By day, he worked at cybersecurity companies and volunteered at a nonprofit that responsibly discloses security vulnerabilities to protect people. By night — at least before his 2023 conviction — he was "Umbreon," a hacker who stole personal data on millions of people and extorted victims for money. He went to prison, said he'd changed, gave interviews about turning his life around, and landed a job as an offensive security lead. Then, Dutch police arrested him again in September 2026 on suspicion of helping ShinyHunters steal data on 6.2 million Dutch people.
Think of it like hiring a locksmith who once did time for burglary. You might trust them — and they might genuinely be trying to go straight. But the skills don't disappear, old networks don't vanish, and the temptation or pressure to slip back can be real. The cybersecurity industry actively recruits people with hacking backgrounds, because understanding how attackers think is genuinely valuable. That creates a tension: the same knowledge that makes someone a great defender also makes them a capable attacker.
For regular people and organizations, this case is a reminder that trust — especially in anyone with privileged access to systems or sensitive data — should come with ongoing accountability, not just a background check at hiring time. Security teams should ask: who in our organization can access what, and are we monitoring that access in a way that would catch misuse? Insider threats, whether from malicious employees or people who were trusted and then went bad, are one of the hardest problems in security precisely because normal defenses are often designed to stop outsiders.
Watch for what comes next here. ShinyHunters responded to van der Stap's arrest by stealing FBI data and extorting a Russian ransomware group — a brazen escalation designed to show they won't be intimidated. Dutch prosecutors face the difficult task of building a case while the group continues to operate and retaliate. Whether van der Stap is convicted again or not, the episode exposes how thin the line between "reformed hacker" and "active criminal" can sometimes be — and how little margin for error there is when the stakes are millions of people's personal data.