Home / News / Article
Dark Reading · 3 min read

‘TheMailer’ Phishing Kit Puts AI in the Driver’s Seat

The commercial kit uses attacker-supplied AI models to personalize lures, evade spam filters, and scale campaigns to tens of thousands of targets.

A new phishing kit is capable of orchestrating most elements of the attack through AI functionality, including tailoring email lures specifically to bypass content filters.

That’s according to new Varonis research detailing “TheMailer,” a phishing kit that leverages large language models (LLM) to a degree hardly seen in the cybercriminal ecosystem. Phishing kits and phishing-as-a-service (PhaaS) platforms have gained significant popularity in recent years, and AI has only made them more capable. To date, phishing kit AI functionality focuses mostly on generating lures, coding realistic phishing websites, conducting reconnaissance, and automating simple repetitive tasks.

TheMailer distinguishes itself in two primary ways. One, it automates much of the phishing process from end to end using the attacker’s chosen AI model. Two, as Varonis senior security researcher Daniel Kelley wrote in the blog post, the kit can tailor lures to each mail provider and “rewrite them until they bypass spam filters.”

How TheMailer Uses LLMs at Every Stage

The attacker’s “customer journey” begins by buying a lifetime license for between $3,000 and $6,000 in cryptocurrency (depending on the deal and which AI functions are included). The customer operator then plugs in their API key, selects an AI model (Claude, ChatGPT, DeepSeek, Grok, Gemini, and Mistral are included), and then they’re more or less ready to start conducting campaigns. Kelley noted that customers only need to bring SMTP servers, proxies, and target email lists.

TheMailer can generate lures from a short prompt or a real brand email, personalize the emails for each recipient (including email address, body, and subject line), remove AI writing patterns, and adjust content for spam filters.

Kelley tells Dark Reading the kit is built to scale against targets.

“The seller’s screenshots show a single campaign of nearly 25,000 emails and a list of more than 145,000 targets being sorted by mail provider. What’s different is that every recipient gets their own version of the email, subject line, and attachment, so a bulk campaign no longer looks like one,” he says. “I wouldn’t call it a spear-phishing tool. The personalization we saw is at the level of a name on an invoice, and we didn’t see features for researching individual targets. It’s opportunistic phishing with the polish of something targeted.”

Brace For the AI Phishing Kit Future

It’s unclear how successful the kit is — attacker LLM tools are historically capable of undertaking tasks at scale, but the success rates of these tasks vary — though Varonis assesses TheMailer to have a large customer base. Moreover, the kit may be a sign of things to come.

“I’d expect more kits to follow,” Kelley says, “because once one seller offers this, buyers start asking for it elsewhere.”

For defenders, Varonis offered a few main recommendations. Because TheMailer is built specifically to get past content filters, organizations should consider email authentication practices like SPF, DKIM, and DMARC enforcement; monitoring for lookalike and new domains; and checking links through full redirect chains at the time of the click.

And when conducting phishing training, Varonis recommends making it reflect how emails look in a post-AI world.

“Bad grammar isn’t a reliable sign anymore, so teach users to be wary of urgency, unexpected requests, and login prompts, especially ‘unusual activity’ or account statement alerts,” the blog post read. Finally, “Least-privilege access and data monitoring limit the damage if someone clicks anyway.”

Originally published on Dark Reading