AI-Powered Hacker Breached 600 FortiGate Firewalls in 5 Weeks

TL;DR: Between January 11 and February 18, 2026, a single Russian-speaking threat actor used commercial AI tools (like ChatGPT) to breach 600+ FortiGate firewalls across 55 countries. No zero-days needed, just exposed management ports and weak passwords. Amazon's threat intelligence team says this campaign proves AI is now lowering the barrier to large-scale cyberattacks. The attacker had "limited technical capabilities" but AI helped automate reconnaissance, tool development, and credential extraction. If you run FortiGate, check your management interface exposure now.

What Happened

Amazon Threat Intelligence revealed on February 21, 2026, that they tracked a financially motivated attacker who compromised over 600 FortiGate firewall appliances across more than 55 countries in just five weeks.^[1]

The attacker used multiple commercial generative AI services throughout the entire operation. Amazon's CISO CJ Moses put it bluntly:

"This activity is distinguished by the threat actor's use of multiple commercial GenAI services to implement and scale well-known attack techniques throughout every phase of their operations, despite their limited technical capabilities."^[1]

No exploits were needed. No zero-days. The attacker found exposed management interfaces and weak single-factor passwords, then let AI do the rest.

How AI Powered the Attack

The threat actor deployed AI as an operational assistant across every phase:^[2]

Attack planning: AI generated step-by-step attack methodologies
Tool development: Custom reconnaissance tools built in Go and Python, written by AI
Credential extraction: AI-generated scripts to parse configurations and extract credentials
Automation: AI helped scale what would normally require a team of hackers
Network pivoting: AI served as an operational assistant for lateral movement tactics

The attacker scanned ports 443, 8443, 10443, and 4443 for vulnerable FortiGate admin panels. Brute-force attacks succeeded against common passwords. Once inside, AI-generated tools extracted SSL-VPN credentials, firewall policies, network topology, and IPsec VPN settings.^[3]

The Damage

600+ FortiGate firewalls compromised. 55+ countries hit. Geographic spread included:^[2]

South Asia and Southeast Asia
Latin America and the Caribbean
West Africa
Northern Europe

What the attacker stole:^[2]

FortiGate device configurations with embedded credentials
Microsoft Active Directory credential databases
VPN connection details and authentication materials
Backup infrastructure access credentials

The campaign also targeted Veeam Backup & Replication servers using custom PowerShell scripts, a classic ransomware precursor. Take out the backups, then encrypt everything.^[3]

What Failed

Basic security hygiene. Amazon identified these failures:^[2]

Exposed management ports: FortiGate admin panels accessible from the internet
Weak passwords: Single-factor authentication with guessable credentials
No MFA: Admin accounts protected only by passwords
Poor network segmentation: Once inside, lateral movement was trivial
Insufficient monitoring: Post-compromise activity went undetected

This wasn't sophisticated. It was opportunistic. AI just made it faster and easier to find and exploit weak targets at scale.

Why This Matters

This is the first major documented case of AI being used to scale a cyberattack campaign across dozens of countries by a low-skilled threat actor.

Security Boulevard called it "a defining demonstration of how AI is lowering the technical entry barrier to offensive cyber operations."^[2]

What used to require a skilled team can now be done by one person with ChatGPT access. AI doesn't write the malware: it automates the reconnaissance, generates the tools, and helps an attacker move faster than defenders can respond. The flip side is the fight over AI safety guardrails, like the one that got Anthropic banned from federal contracts.

The threat actor was described as having "limited technical capabilities."^[1] But capability didn't matter. AI filled the gap.

What To Do Right Now

Lock Down Management Interfaces

FortiGate admin panels should never be internet-accessible. Use VPN-only access or restrict to trusted internal IPs.

Enable MFA Everywhere

Single-factor authentication on firewall admin accounts is asking to be compromised. Enable MFA immediately.

Audit Admin Accounts

Check for unauthorized accounts. Look for generic names like "helpdesk," "support," or "backup" that you didn't create.

Rotate Credentials

If your FortiGate config was exfiltrated, every credential stored in it is compromised. Rotate AD, VPN, and backup passwords.

Amazon's recommendations:^[2]

Maintain aggressive patch management, especially for edge devices
Strengthen credential protections across all network equipment
Implement proper network segmentation
Enhance detection systems for post-compromise indicators

The Bigger Picture

This comes just one month after another FortiGate exploitation wave in January 2026, where attackers bypassed authentication even on "fully patched" devices.^[4] The same Chinese state-backed playbook surfaced when CISA found a backdoor buried inside a federal Cisco firewall that survived reboots.

FortiGate firewalls remain a prime target. They sit at network boundaries, contain credentials for everything behind them, and when compromised, give attackers keys to the kingdom.

Now AI is making these attacks easier to scale. The barrier isn't technical skill anymore: it's finding weak targets. And AI helps with that too.

The Bottom Line

One attacker. Multiple AI tools. Five weeks. 600+ firewalls across 55 countries.

This is what cybercrime looks like in 2026. The attackers don't need to be skilled: they just need AI access and targets who haven't done the basics.

Check your FortiGate management interface exposure. Enable MFA. Assume the next attacker will have AI helping them move faster than you.