AI for Good – Cybersecurity Good

AI can be amazing. Sometimes. And it can be bad sometimes. This time it is good. And bad.

An AI-assisted hacker, Russian speaking, used multiple generative AI services to breach more than 600 FortiGate firewalls across 55 countries in 5 weeks.

Should I blame AI for this? Maybe not, but the Russian used AI to power his hacking campaign.

He was able to breach these 600 firewalls because (A) they exposed their management interfaces to the Internet (it is possible to do this safely, but many people do not), (B) they used weak userids and passwords and (C) they did not turn on MFA. Note to readers: make sure that you are not guilty of this.

The hacker discovered the firewalls and then used brute force to try common passwords.

The hacker, once in, extracted the device’s configuration and used AI-assisted Python and Go to parse and decrypt the files. Once connected via the VPN the hacker used custom reconnaissance tools to scan the network.

The code was likely AI generated, Amazon (who discovered it) thinks.

The hacker also went after Veeam backup servers with the same mindset.

If one firewall or Veeam server was not easily exploitable, the hacker just went on to another.

After all of this, the attacker fed the data gained into ARXON which queried large language models to generate a structured attack plan.

Pretty amazing, that AI stuff. More details, if you are curious, at the link.

EXPECT A LOT MORE OF THIS BECAUSE IT IS DOWN RIGHT EASY. Credit: Bleeping Computer

On the other side of AI automation, Claude Code Security discovered over 500 vulnerabilities and the tool is now available to you and me.

Anthropic pointed its most advanced AI model, Claude Opus 4.6, at production open-source codebases. They found a bucket of security holes. They found more than 500 high-security vulnerabilities, some of which had survived decades of expert review and millions of hours of fuzzing.

Now Claude Code Security is available commercially. Credit: Venture Beat

by