I’m reading national-security AI from Mexico, and the newest pitch isn’t a chatbot — it’s undercover software. Axios got the exclusive: Washington, D.C. startup Aslan is publicly launching after a $20.8 million seed to run AI agents inside criminal forums, Telegram channels, and other underground spaces for U.S. law enforcement and the intel community.
FinSMES lists the same figure under the name Aslan Protects — co-led by Khosla Ventures and XYZ Venture Capital, with BoxGroup, 2048 Ventures, Liquid 2 Ventures, Alumni Ventures, and other defense-tech backers. Valuation wasn’t disclosed. I’m not inventing one.
A harness, not another scraper
Per Axios, Aslan sells a harness that coordinates the models agencies already use — so agents can collect digital evidence where human infiltrators are too slow, too expensive, or too exposed. CEO Chase Reid told Axios the agents are human-supervised and behave like FBI analysts or undercover spies in those ecosystems. “It’s not just passively collecting,” he said. “Our agents are capable of actively engaging and doing cyber effects within these online ecosystems.”
AI Weekly (summarizing Axios) adds that Reid founded the company with Ansel Tessier after Bain Capital and JP Morgan. Founded in 2025. Team size around ten people before the expansion hire plan.
21 days on Telegram — and a hard red line
Axios says agencies have mostly run short evaluations while they figure out guardrails, data collection, and storage. The FBI, Homeland Security Investigations, and other shops have already worked with Aslan on a mix of investigations, Reid said. In one case he described, agents ran autonomously for 21 days and two of them infiltrated “the small, small slice of Telegram where six people were coordinating smuggling,” pulling who the coordinators were and when a border crossing would happen. AI Weekly and Pomegra repeat that 21-day claim from the same Axios package.
Company-claimed wins relayed by Axios include mapping a live U.S.–Mexico border smuggling network, a sanctions-evading cyber-fraud marketplace, tech-transfer paths into U.S. AI infrastructure, and a coordinated Chinese recruitment effort aimed at U.S. defense professionals. Treat those as company assertions via the Exclusive — not independent court findings.
Reid’s red line matters from where I sit: he told Axios Aslan won’t run domestic ops that target or surveil Americans, and the startup is focused on the U.S. market for now. “We’re not trying to replace any human,” he said. “This is just scaling existing human judgment beyond what a hiring office can do.” The new money goes to technical hiring, including forward-deployed engineers who can sit with agencies on specific missions.
From Mexico, my take is simple: agentic AI left the demo reel and landed in undercover ops — with human babysitting claimed as the control plane. Watch whether short evals become lasting contracts. Sources: Axios, FinSMES, AI Weekly, Pomegra.
Hero image: green digital code cascade by Markus Spiske on Unsplash (Unsplash License). Cropped, graded, and lightly grained by Tech & AI Pulse. Face-free code still — no people, no logos.
