Glossary
November 16, 2026

Google Finds Hackers' AI Programs Breaking Into Websites With No Person Involved, as the FBI Warns About a Widely Used AI Hacking Tool

Google's security researchers say they have found, for the first time, criminals' AI programs finding a weakness in website software, writing an attack and carrying it out with no person involved. On the same day, an FBI warning came to light about ARTEX, a free AI hacking tool it says has been used against organizations around the world, sometimes getting deep inside in under three hours.

Two warnings today describe the same shift: break-ins that once needed skilled people are increasingly being run by AI programs.

The first comes from Google's threat intelligence team, which tracks hackers. It says it found an unprotected computer belonging to an attacker, probably Russian-speaking judging by notes left in the code, holding a set of AI programs that worked as a team. When makers of popular website software released a security fix, the programs compared the software before and after the fix to work out what weakness had been repaired. Then they wrote an attack aimed at websites that had not yet installed the fix, and launched it. Google says no person was needed at any step, and that this is the first time it has seen that happen. The computer also held data stolen from victims, mostly French media and entertainment websites and schools and universities. Google did not say how many were affected.

The second is a warning from the FBI, dated November 13 and reported today by the tech news site TechCrunch. It concerns ARTEX, a free AI hacking tool that was posted publicly in July. A security company reported last month that a China-linked spying group was using it. Its author said it was meant for legitimate security testing and took it down in October, but the FBI says copies, altered versions and a paid version keep it widely available. ARTEX hands a goal, written in plain language, to a team of AI programs, which then hunt for weak spots and work their way in. They can run on chatbots from Anthropic, OpenAI and others, though the FBI says cheap Chinese ones, mostly DeepSeek's, are the most common.

The FBI says that since at least August, different groups have used ARTEX against banks, government offices, tech, defense, energy and health care companies, and law firms. Some break-ins went from first look to full control of a computer system in under three hours. Personal information on about 67,000 people was taken from financial companies in South Korea. In another case, attackers went through ConnectWise, a company that backs up other businesses' data, and tried to sign in to more than 2,250 of its customers' accounts. ConnectWise told TechCrunch it cannot confirm that number. Researchers have found more than 1,600 ARTEX control panels sitting openly on the internet.

There are reasons for some calm. Most of these attacks get in through ordinary weak spots, such as passwords that were left exposed or stolen, or accounts without a second login check, so the usual advice still works. Google says these tools are noisy, leaving obvious traces that security software can catch, and that many attacks written by AI simply don't work. The worry is speed. "Defenders will often arrive after the attacker has finished," Arve Kjoelen, security chief at the security company Barracuda, which was itself broken into, told TechCrunch. Google expects these tools to keep spreading and to need less and less human help.

Sources

  1. Google Threat Intelligence Group — AI threat tracker: agentic threat activity crosses a significant thresholdcloud.google.com · primary
  2. FBI — FLASH warning on the ARTEX AI hacking tool (PDF)ic3.gov · primary
  3. TechCrunch — FBI says ARTEX AI hacking tool used to breach networks in under three hourstechcrunch.com
  4. The Register — Google says an AI agent turned a security patch into a working exploit with no human helptheregister.com
Follow the timeline
Get an email when new entries are added.
© 2026 Sugarpine