Hackers Spied on a Stock Exchange Executive’s Outlook Mailbox for Five Months
Unknown attackers spent at least five months quietly inside the Outlook mailbox of a senior executive at a major global stock exchange, exfiltrating the inbox in small, repeated batches and routing the stolen data through Dropbox and OneDrive so the traffic blended in with normal cloud activity. The campaign points to espionage rather than financial gain — the executive’s inbox would have contained non-public listing details, enforcement matters, deal terms, and market-moving plans. Neither the executive nor the exchange was identified; the first malicious activity was observed as far back as October 10, 2025.
VS Code Vulnerability Allows One-Click GitHub Token Theft
A security researcher publicly disclosed a severe Visual Studio Code zero-day — along with a working proof-of-concept — without advance notice to Microsoft, citing a prior bad experience with the company’s disclosure process. The attack involves a specially crafted Jupyter notebook that, when opened on github.dev, silently simulates keystrokes to install a malicious extension that steals the victim’s GitHub OAuth token, granting full read and write access to all of their repositories. Microsoft pushed a fix on June 3 for the browser-based version, but the desktop version of VS Code remains unpatched.
CISA Warns of Cyberattacks Targeting Fuel Tank Monitoring Systems
A joint advisory from CISA, the FBI, the NSA, the Department of Energy, and several other U.S. government agencies warns that hackers are actively targeting internet-exposed automatic tank gauge (ATG) systems used to monitor fuel and liquid storage tanks at gas stations, airports, military bases, and other critical infrastructure. ATG systems are often left internet-accessible with default credentials, and successful attacks could allow adversaries to manipulate fuel levels, disable alarms, or cause physical damage to storage infrastructure.
Rust-Written IronWorm Hits NPM Supply Chain
A newly identified supply chain campaign tracked as IronWorm has compromised npm packages to steal developer credentials and then uses those same credentials to propagate further across the software supply chain — a self-replicating pattern reminiscent of earlier Shai-Hulud attacks. The worm is written in Rust and targets developer machines, harvesting tokens and secrets before moving laterally into other repositories and pipelines that the compromised accounts have access to.
Autonomous AI-Driven Worm Can Reason Its Way Through Corporate Networks
Researchers from the University of Toronto, the Vector Institute, and the University of Cambridge have built and tested a proof-of-concept AI-driven worm that foregoes a fixed exploit list and instead reasons about each target on the fly, generating attack strategies in real time using a small open-weight language model it runs locally on machines it has already compromised. In a 33-host isolated test network, the worm correctly identified 31 vulnerabilities, exploited 23 hosts, and propagated to 20 — and it demonstrated the ability to adapt to vulnerabilities disclosed after its model’s training cutoff by reading public advisories at runtime.
