Hackers Turn AI Agent Into a Cyber Weapon After Deleting Its Safety Refusals
Researchers exposed BlackHatSect0r && DXQRTXX infrastructure showing a safety-disabled Hermes AI agent used to automate scanning, credential harvesting, and vishing against French telecom subscribers.
Socradar-analyzed infrastructure attributed to French-speaking crew BlackHatSect0r && DXQRTXX exposed 4.9 GB across 9,299 files, including the DXSCAN Go-based C2 platform, 16,834 harvested credentials, and a database of nearly 450,000 records used for a vishing campaign targeting older French telecom subscribers with Societe Generale-themed lures. The crew ran a self-hosted Nous Research Hermes agent on a DeepSeek model with refusal instructions removed and HERMES_DISABLE_SAFETY=1 set, using it for scanning, secret hunting, and Telegram reporting. DXSCAN queued 2.75 million domains and reached over 726,000 hosts; the toolkit relied on exposed secrets and misconfigurations rather than novel exploitation.
- Exposed server held 4.9 GB across 9,299 files, including DXSCAN C2 platform and 16,834 harvested credentials
- Hermes agent ran on DeepSeek with safety refusal removed and HERMES_DISABLE_SAFETY=1 set
- DXSCAN queued 2.75 million domains and reached over 726,000 hosts, scanning 200+ secret patterns
- Vishing campaign targeted 1,697 French subscribers aged 50+ with Societe Generale-themed fraud lures
- JWT-forging tooling targeted a cryptocurrency exchange that used the literal signing secret 'secret'
- CVEs
- CVE-2026-42530
- ATT&CK
- T1595 Active ScanningT1552.001 Unsecured Credentials: Credentials In FilesT1566.004 Phishing: Spearphishing VoiceT1102 Web ServiceT1606.001 Forge Web Credentials: Web Tokens
- Vendors
- Nous ResearchNGINX
- Products
- DXSCANHermes agent
- Threat actors
- BlackHatSect0r && DXQRTXX
- Organizations
- Socradar
- Countries
- FranceUnited States
Vulnerabilities mentioned
All| CVE | Vulnerability | CVSS | EPSS | Signals | Published |
|---|---|---|---|---|---|
| CVE-2026-42530 | NGINX Open Source has a vulnerability in the ngx_http_v3_module module | 9.2 | 6% | PoC |
From the original article
4 min read at gbhackers.com
A French-speaking cybercrime crew calling itself BlackHatSect0r && DXQRTXX allegedly disabled safety controls in a self-hosted AI agent and used the resulting system to automate mass credential harvesting, target discovery, phishing preparation, and attack orchestration. The internet-exposed server reportedly contained 4.
By Mayura Kathir · gbhackers.com
A short excerpt; the article and its images belong to the publisher. Summary, key facts and assessment above are AI-written from the article and can be wrong.