The FBI IC3 issued a series of alerts covering various cyber threats from May through August 2026. Iranian-Affiliated Cyber Actors exploited Programmable Logic Controllers across US critical infrastructure on July 22 and April 7. Russian State-Supported Cyber Actors conducted phishing campaigns targeting Zimbra Collaboration Suite users on July 23. North Korean Kimsuky Actors used malicious QR codes in spearphishing efforts aimed at U.S. entities on January 8. Gunra Ransomware was flagged on August 10, and the Silent Ransom Group impersonated IT personnel through social engineering on May 26. TeamPCP, a cyber criminal group, was reported on July 2. CISA urged hardening Automatic Tank Gauge Systems on June 2, and advised isolating vital systems via CI Fortify on July 28. The Government of Iran Cyber Actors deployed Telegram Command-and-Control to push malware on March 20. SocksEscort used AVrecon-infected routers as residential proxies on March 12. ATM jackpotting incidents increased in February, with compromised devices exploited for financial gain. North Korean IT Workers were also noted in an alert issued on July 31. These alerts reflect a broad spectrum of threat actors and attack vectors affecting both enterprise and operational technology environments.

Between July 1 and July 4, 2026, a China-linked operator leveraged autonomous AI tools to target Taiwanese government infrastructure. The campaign focused on Windows Ancillary Function Driver for WinSock, Windows User Profile Service, and Windows Container Isolation FS Filter Driver (unionfs.sys). Tenable’s Research Special Operations team tracked the activity as part of a broader cluster of agentic AI operations. The attacker mapped 21 connected systems, compromised 85 accounts, and exfiltrated more than 2,564 personnel records. The agents used Hermes Agent and OpenClaw frameworks with Bayesian decision engines to coordinate up to eight parallel sub-agents. They scraped authentication metadata from a government portal, then accessed a GitBook documentation site hosting the national SSO integration guide. The AI agents generated password variations and solved CAPTCHA challenges automatically. No single CVE drove the breach; instead, misconfigurations, exposed interfaces, and weak credentials were exploited dynamically. The operation expanded to include Taiwan’s national nuclear safety agency, seven energy companies, government IT vendors, and a government email system. The attack was confirmed by Taiwan’s Ministry of Digital Affairs on August 13, 2026.

On August 4, 2026, attackers gained access to the GitHub account of a maintainer and injected malicious code into keyv and eight related npm packages. The compromised packages include keyv 6.0.0, flat-cache 6.1.24, file-entry-cache 11.1.6, and others with combined monthly downloads exceeding 2 billion. Each affected package contained two new files: setup.mjs and Math_Symbol.js. The setup.mjs file is a heavily obfuscated dropper that silently downloads the Bun JavaScript runtime from github.com/oven-sh/bun/releases/download/bun-v1.3.13/ and uses it to execute Math_Symbol.js. The Math_Symbol.js payload is a 728 KB JavaScript file designed to harvest credentials from victim environments, including npm tokens, GitHub tokens, AWS credentials, Kubernetes secrets, HashiCorp Vault tokens, Stripe and Slack tokens, and more. It also includes worm-like propagation functionality to infect other packages installed by maintainers who had installed one of the compromised packages. The payload exfiltrates harvested secrets to a public GitHub repository with the description 'Shai-Hulud: Here We Go Again'.