Varonis Threat Labs has identified TheMailer, a self-hosted phishing platform that leverages commercial AI models to dynamically generate and rewrite email templates. Unlike static phishing kits, TheMailer integrates an Inbox Loop feature that sends test messages to an IMAP mailbox, checks delivery status, and automatically requests rewrites from the LLM if the message lands in spam. This represents a shift from manual template tweaking to automated, feedback-driven content optimization for threat actors.
Operator-Supplied Infrastructure and Model Keys
TheMailer operates on a bring-your-own-infrastructure model. Operators must supply their own API keys for major commercial models including ChatGPT, Claude, Gemini, Grok, Mistral, and DeepSeek. Additionally, users provide their own email sending systems, middleware, and recipient lists. This design allows attackers to distribute costs and avoid single points of failure associated with centralized phishing services, while giving them granular control over which LLM generates the content for specific campaigns.
Automated Spam Avoidance and Provider-Specific Logic
The platformβs core technical advantage lies in its ability to control message placement. By checking IMAP folders, TheMailer determines if a test email reached the inbox or spam. If flagged, it triggers a rewrite cycle. Furthermore, TheMailer groups recipients by providerβsuch as Gmail, Microsoft, GMX, Yandex, and French servicesβand applies different sending limits and guidelines for each. This provider-specific logic helps bypass rate limits and heuristic filters that vary significantly across major email ecosystems.
Licensing Costs and Target Data Analysis
Licenses for TheMailer are advertised between $3.000 and $6.000, depending on the package and included AI capabilities. While the toolβs advertisement garnered over 24.000 views, actual adoption numbers remain unclear. A screenshot of the platform displays a target list of 145.133 entries, primarily from Japanese business and academic sectors. However, Varonis notes that the presence of these targets in the tool does not confirm that these organizations have been successfully attacked or compromised.
Key Takeaways
- TheMailer uses an IMAP feedback loop to automatically rewrite emails that land in spam.
- Operators must bring their own API keys for models like GPT, Claude, and Gemini.
- The tool applies provider-specific sending rules for Gmail, Microsoft, and others.
- Licenses cost $3.000 to $6.000, with a reported target list of 145.133 entries.
The Bottom Line
Phishing is becoming a DevOps problem; attackers are now using CI/CD-style feedback loops to optimize social engineering payloads. The requirement for operators to bring their own API keys suggests this is a professionalized, high-budget operation rather than a script-kiddie toy.