The Daily Byte
Breaking
Welcome to The Daily Byte — real stories from across Pakistan.
Saturday, 21 November 2026 · Pakistan

The Daily Byte

Stories from Pakistan that deserve to be heard

FBI warns AI tool Artex automates cyberattacks

By The Daily Byte ·14 Nov 2026 ·3 min read
The FBI's J. Edgar Hoover Building headquarters in Washington DC

The FBI has warned that an artificial intelligence hacking framework called Artex is letting attackers with limited technical skill run complex intrusions at machine speed, with some break-ins reaching privileged access in under three hours. The bureau issued the warning in a FLASH intelligence bulletin, FLASH-20261113-001, on 13 November 2026, urging organisations across ten major sectors to follow the mitigation steps it sets out.

Artex is an open-source, large language model-driven penetration testing framework. It takes a natural-language goal from an operator and turns it into machine commands, chaining together the stages of an attack — finding a weakness, breaking in, stealing credentials and moving deeper into a network — with little human input. The FBI said it "significantly reduces the time required to chain attack techniques, with some intrusions progressing from initial targeting to privileged access in under three hours."

What Artex does

Artex does not contain its own AI model. Instead, it plugs into outside models, including OpenAI's GPT, Anthropic's Claude, DeepSeek, Grok, GLM and Kimi, and uses autonomous software agents to carry out much of the work. That design lets a single operator run simultaneous attacks against many targets, and lets people with minimal technical expertise attempt intrusions that once needed a skilled team.

The framework first appeared on GitHub in the summer of 2026 under the developer name Autumn-27, and has been attributed to Chinese cybersecurity engineer Li Puhua. The repository was later removed after the developer said the platform was being abused, but it continues to spread through community forks, local copies and commercial variants. Research firm Censys, cited in the FBI bulletin, has identified more than 1,600 publicly exposed Artex console installations, with private deployments likely pushing the total higher.

Who is being targeted

The FBI said malicious actors have used Artex since at least August 2026 in campaigns against organisations worldwide in financial services, government, technology, defence, semiconductors, energy, healthcare, manufacturing, telecommunications and legal services. Cryptocurrency exchanges are also among the targets.

The bulletin describes real campaigns. A South Korean financial-sector campaign exposed personal information belonging to about 67,000 people, and a supply-chain compromise of ConnectWise cloud backup software affected more than 2,250 downstream customer environments. Microsoft reported on 21 October that a China-aligned threat actor known as Storm-0940 used Artex to automate attacks against hundreds of organisations worldwide, including the ConnectWise incident.

Amazon's threat intelligence team also reported on 13 November that its MadPot decoy sensors had captured AI agent requests carrying traces of offensive campaigns run with Artex between 3 October and 5 November, including large-scale credential theft used to fund further AI model use and operations against cryptocurrency platforms using real-time deepfake injection to bypass identity checks.

What organisations should do

The FBI's FLASH bulletin sets out tactics, techniques and procedures and indicators of compromise associated with Artex, and the bureau said organisations should follow the mitigation steps in it to improve their defences against Artex and similar agentic attack frameworks. The warning, issued by the FBI Cyber Division under Assistant Director Brett Leatherman, is the bureau's most detailed public statement yet on a tool that security researchers say is compressing the time defenders have to spot and stop an intrusion. Read the FBI's warning as reported by GazeWire and the Record's report.