A free Chinese AI hacking tool called ARTEX can take an attacker from first probe to admin-level control of a company network in under three hours, and the FBI says it is already being used against banks, governments and defense firms around the world.

Key Takeaways

  • ARTEX turns plain-English goals like “gain admin panel access” into automated attacks run by a team of AI agents

  • The FBI says some ARTEX intrusions reached privileged access in under three hours, and it lets attackers with little skill run complex operations

  • More than 1,600 ARTEX control panels were found exposed online in early November, and the real number is likely far higher

  • Viral posts claim anyone with an iPhone can now launch cyberattacks, but the FBI’s warning says nothing of the sort

In a FLASH advisory dated Nov. 13, the FBI said that since at least August 2026, “a variety of malicious cyber actors” have used ARTEX against financial services, government, technology, defense, semiconductor, energy, healthcare, manufacturing, telecommunications and legal organizations. The bureau said the tool “enables threat actors with minimal technical expertise to conduct complex, multi-target operations simultaneously.”

The warning is spreading fast online this week, with posts claiming the tool lets anyone with an iPhone hack companies. The real story is less cinematic but more worrying for businesses.

What ARTEX Actually Does

ARTEX was released as open-source software on GitHub on July 26, 2026, by a developer using the name Autumn-27, according to internet scanning firm Censys. It was pitched as a penetration testing tool, the kind security teams use to attack their own systems and find holes before criminals do.

The user types a goal and a target. A “planner” AI breaks the goal into steps, and “worker” AI agents run scans, try passwords, exploit weak spots and report back, looping until the job is done. ARTEX does not have its own AI brain. It plugs into large language models (LLMs) from outside providers. The FBI said attackers’ copies mostly run on Chinese models, with DeepSeek’s cheaper models dominating “due to their balance of capabilities, relatively low cost, and lack of observable guardrails.”

Speed is the big change. Google‘s Threat Intelligence Group said it watched ARTEX fire 27 attack events in 23 seconds during one intrusion, in bursts no human operator could match.

ARTEX By The NumbersFigureWho Reported ItTime from first target to privileged accessUnder 3 hours in some attacksFBIExposed ARTEX control panels onlineMore than 1,600FBI, citing researchersCompany accounts hit with sign-in attempts after the ConnectWise breachMore than 2,250FBIPeople whose data was taken from South Korean finance firmsAbout 67,000FBIAttack actions in one burst27 in 23 secondsGoogle Threat Intelligence GroupGitHub stars in its first 10 weeks1,500 to 2,000FBI, Censys

Sources: FBI FLASH advisory, Nov. 13, 2026, Google Threat Intelligence Group and Censys research.

Who Has Used It

The FBI linked ARTEX to two major campaigns. Financially motivated hackers used it against South Korean financial institutions, stealing personal information on about 67,000 people. A China-aligned group that Microsoft tracks as Storm-0940 used it in a supply chain attack on ConnectWise Cloud Backup customers, which led to attempted sign-ins at more than 2,250 downstream company accounts.

Google said it has also seen extortion gangs and other suspected China-linked groups use ARTEX inside victims’ networks, hitting government, technology, chemicals and manufacturing targets in the U.S.

The developer pulled the GitHub project on Oct. 8, saying it was meant only for authorized testing. It did not matter. The FBI said ARTEX keeps spreading through copies, translated versions and a paid version called ARTEX PRO.

Also Read: Hackers Leak JPMorgan Customer Data After Breaking Into Its Vendors

What About The iPhone Claim?

The FBI advisory does not say iPhone owners can launch attacks. ARTEX is controlled through a web page, so an operator could open that page on a phone’s browser, which is likely where the iPhone talk comes from. But someone still needs a server running ARTEX, paid access to an AI model and, in many cases, stolen passwords for the target. Google said attackers often fed ARTEX compromised credentials and then let it run the rest of the attack.

The one iPhone link in the FBI’s warning is different and more serious for app makers. The bureau said ARTEX agents have changed iOS app distribution systems by abusing exposed cloud credentials, a supply chain risk that could reach the apps people download.

Why This Matters

ARTEX does not invent new hacking tricks. It runs the old ones, like weak passwords, unpatched servers and leaked keys, at machine speed. That shrinks the time defenders have to spot an attack and react. On X, the FBI said it wants Americans to protect themselves against criminals and state actors using AI “to run familiar techniques at machine speed.”

It also puts AI companies in the middle. ARTEX works with Anthropic’s Claude and OpenAI’s GPT models as well as Chinese models, so the companies selling AI access are now part of the fight over who gets cut off. Cheap models with few guardrails are winning with attackers, according to the FBI.

What To Watch Next

Expect more warnings as security firms keep finding ARTEX copies and rebrands online. The commercial ARTEX PRO version means the tool now has a business model, which usually means steady updates.

For companies, the fix is the boring stuff done fast. The FBI urged organizations to lock down access controls, watch for bursts of automated activity and report suspected breaches to its Internet Crime Complaint Center. Google’s advice includes short-lived passwords and keys, giving accounts only the access they need and separating critical systems, steps that slow down an attacker that never sleeps.

Read Next: Google Treats Its Own AI Agents Like Rogue Staff After Catching Cheats