110 Million Credentials Stolen From Your Firewall

A financially motivated Russian-speaking group has been running one of the most methodical credential-harvesting operations seen this year, according to SOCRadar's full report via The Hacker News. The campaign, active since February 2026, has compromised over 430,000 FortiGate firewalls worldwide.

The attack follows a ruthless five-step playbook. First, the group scanned the internet for exposed FortiGate devices using tools like Masscan and Shodan. Then, they forced their way in through credential stuffing (trying stolen username and password combinations across many targets) and dictionary attacks against admin panels and VPN portals.

Once inside, they deployed a Golang-based (Google's programming language, often used to build fast, cross-platform tools) tool called FortigateSniffer. It exploits a built-in FortiOS diagnostic command to passively intercept live authentication traffic, like slipping a tap onto a telephone wire. The sniffer captured credentials across 24 protocols — everything from LDAP to RDP to MySQL.

The haul was staggering in its precision: 110 million credentials total, including 14.8 million RADIUS credentials, 924,000 NTLM hashes, 130,000 Kerberos hashes, and 89 million MySQL authentication tokens. Password hashes were cracked and fed into lateral movement (moving deeper through a network after initial access) across Active Directory environments.

Small and medium businesses with under 200 employees were the primary targets, particularly in IT services, the US, and India. Breaching IT service providers gives attackers a shortcut into their clients' systems.

What to do: If your organisation uses FortiGate devices, check for unrecognised SSH sessions or unexpected sniffer processes running on your firewall. Rotate all administrative credentials immediately. Enable multi-factor authentication on every exposed admin panel and VPN portal.

Security firm AIR built a malicious AI agent skill (a bundle of instructions that an AI agent loads and executes with roughly the same authority as a direct user command), published it to a marketplace, ran an Instagram ad, and watched it reach approximately 26,000 agents — including corporate ones. Every scanner it was tested against rated it clean, according to The Hacker News.

The trick was structural. Scanners examine what ships inside the skill package. AIR's skill contained nothing suspicious — it pointed to an external URL for setup instructions. That URL was clean at scan time. After widespread installation, AIR swapped the page to deliver a script instead.

The payload in this case only collected email addresses. A real attacker could read files, exfiltrate data, or pivot into internal systems.

What to do: Treat AI agent skills like software installs, not app store downloads. Vet every external URL a skill references, and restrict agent permissions to the minimum required for the task.

What Cisco Unified CM is: Cisco Unified CM is an enterprise on-premises platform that manages voice calls, video conferencing, and internal communications for large organisations — the phone system backbone for thousands of businesses.

What it is: A critical flaw in the WebDialer service component allows a remote attacker with no credentials to perform SSRF (server-side request forgery — when an attacker tricks a server into making requests on their behalf), write arbitrary files to the underlying operating system, and escalate privileges to root. The WebDialer service must be enabled for exploitation, and it is disabled by default.

Who's at risk: Any enterprise running Cisco Unified CM with the WebDialer service enabled and internet-accessible. Given Unified CM's widespread use, both financially motivated attackers and state-sponsored groups have strong incentive to exploit this.

CVSS: Not yet officially scored in public sources — treat as Critical until confirmed. CVE ID not confirmed in source as scored — check the official Cisco advisory before acting.

Root cause: The WebDialer service fails to properly validate or restrict incoming HTTP requests. It processes attacker-supplied input without checking whether the request originates from a legitimate internal source. This allows the server to be directed toward resources it should never reach.

Attack vector: An attacker sends a crafted HTTP request to the WebDialer endpoint, requiring no login. The server processes it as trusted, allowing the attacker to write a file to the OS and escalate to root. A public proof-of-concept (PoC) exploit is now available, and active exploitation has been observed from at least one source.

Detection strategies:

• Look for unexpected file creation events in OS-level logs on Unified CM nodes
• Monitor WebDialer service logs for requests containing file:// URI patterns
• Check for new or unrecognised processes running with root privileges on Unified CM hosts

Recommended actions:

1. Apply Cisco's patch, released June 3 — prioritise this immediately
2. If patching is not possible today, disable the WebDialer service until the patch is applied
3. Restrict network access to Unified CM admin interfaces to trusted IP ranges only

GitHub has updated its actions/checkout tool — the official action that pulls repository code into automated build pipelines — to block a class of attacks called pwn requests, according to The Hacker News.

A pwn request attack works like letting a stranger into your office to drop off a delivery, and then discovering they rewired the server room while no one was watching. Attackers submit a pull request containing malicious code; a misconfigured workflow runs it automatically with full access to secrets and tokens.

As of June 18, actions/checkout v7 refuses to check out unreviewed fork code in the most dangerous workflow configurations by default.

What to do: If you maintain any GitHub Actions workflows, update actions/checkout to v7 now. Review any workflows using the pull_request_target trigger and confirm they do not check out code from untrusted forks. The full backport to older supported versions lands July 16.

President Trump signed an executive order on June 22 requiring federal agencies to migrate high-value systems to post-quantum cryptography (encryption designed to resist attacks from future quantum computers) by December 31, 2030, per The Hacker News. The urgency is real: adversaries are already collecting encrypted data today to decrypt it once quantum computing matures — a strategy known as "harvest now, decrypt later." Federal contractors should expect compliance requirements to follow. For everyone else, this signals that encryption standards are about to change at scale.