Pomegra
A Washington-area startup has raised $20.8 million to deploy AI agents that go undercover in criminal forums, Telegram channels, and dark-web markets - with the FBI and Homeland Security Investigations already as clients.
Key Takeaways
- Aslan raised $20.8M led by Khosla Ventures and XYZ VC, with 2048 Ventures, BoxGroup, Liquid2, and Alumni Ventures participating.
- AI agents are human-supervised and have been used by FBI and HSI to map smuggling networks and cyber-fraud operations.
- CEO Chase Reid says the system is not designed for domestic surveillance of U.S. citizens.
Lead
Aslan, a national-security AI startup founded in 2025, emerged from stealth on September 1, 2026 after closing a $20.8 million round led by Khosla Ventures and XYZ Venture Capital. The company builds AI software that can be tasked to infiltrate and operate inside criminal forums, Telegram channels, and dark-web environments on behalf of U.S. law enforcement and intelligence agencies. The FBI and Homeland Security Investigations have already run live deployments. The company currently has roughly 10 employees and plans to nearly double headcount by the end of October.
What Does Aslan's Software Actually Do?
The system places human-supervised AI agents into online environments that human undercover officers either cannot access at scale or cannot inhabit without risk of exposure. Aslan's agents can join forums, engage in channel activity, and collect intelligence without a human officer directly present in the environment at each moment. Human operators maintain oversight and control throughout each mission.
Live deployments disclosed by the company have mapped a transnational smuggling operation, investigated sanctions-evading cyber fraud, and identified networks involved in moving U.S. AI infrastructure and technology to China. The company's name - a reference to the lion from C.S. Lewis - is perhaps a nod to the predatory posture the product takes inside criminal ecosystems.
Why Are Khosla and XYZ VC Betting on This Space Now?
The defense and national-security AI sector has seen a measurable shift in investor appetite since 2024. Palantir's expansion into AI-driven operations and the political attention on immigration enforcement, fentanyl trafficking, and sanctions evasion have created a clear funding lane for startups willing to work directly with federal agencies. For Khosla Ventures, a firm that built its identity on frontier technical bets, underwriting a startup that sits at the intersection of large language models and covert law enforcement operations fits a pattern of high-conviction, long-duration positions.
The $20.8 million figure is notable for a company at this stage. It suggests the lead investors see a fast path to recurring government contracts, which carry a different risk profile than consumer subscription revenue. Government clients commit slowly but churn slowly; the question for Aslan is how quickly it can convert early agency relationships into long-term, scaled contracts.
Regulatory and Civil Liberties Context
The product's description will inevitably draw scrutiny from civil liberties advocates and legal scholars focused on Fourth Amendment boundaries around AI-driven surveillance. CEO Chase Reid has preemptively addressed this: the system is not intended for domestic operations targeting U.S. citizens, and human supervision is embedded in the architecture, not bolted on as policy.
That framing matters commercially. Federal agencies deploying AI tools in operational contexts must navigate Inspector General oversight, congressional reporting requirements, and, increasingly, internal AI-use policies mandated by executive order. A system built with human-in-the-loop controls from the ground up is more likely to clear legal review than one retrofitted for compliance after the fact. Whether that architecture holds under real operational pressure is a different question.
What Comes Next for Aslan?
The company will use the capital to expand its technical team and hire engineers capable of working embedded at agency and law enforcement facilities during active missions. That hiring profile - engineers with security clearances who can be physically present at federal sites - is rare and expensive. The constraint on growth is not capital; it is cleared talent.
The longer arc of the business depends on whether AI-driven undercover tooling becomes standard tradecraft across federal law enforcement or remains a niche capability available to a handful of agencies. The fentanyl and human trafficking mandates driving HSI priorities, combined with the focus on sanctions enforcement related to China technology transfers, point toward sustained demand. Multiple agencies pursuing the same criminal networks independently creates an obvious opening for a platform that standardizes the approach.
The civil-liberties risk cuts the other direction: a single high-profile incident involving the misuse of autonomous agents in a law enforcement context could freeze adoption across the federal government and trigger legislation that raises the compliance cost substantially.
Outlook
Aslan's exit from stealth puts a name and a funding total on a category that has existed informally - AI-assisted undercover operations - but lacked a dedicated commercial vendor. The $20.8 million round, backed by top-tier venture capital, signals that at least some investors believe this can become a defensible government technology business rather than a one-agency contract. The company's near-term test is straightforward: convert existing FBI and HSI deployments into multi-year contracts before competitors or in-house agency efforts catch up.
