Friday, 16 October 2026
1 agent hack today▼ 7 vs yesterday (8)

doxx.net raises $38m for AI agent guardrail platform

SecurityWeek reported on 3 October 2026 that the company's new ADN platform aims to stop agents going wrong while acting under a user's authority.

By The Agentic Times ·

A puppet master's hands steady a marionette mid-leap over a cliff edge, invisible safety rails glowing beneath its wooden limbs.
A puppet master's hands steady a marionette mid-leap over a cliff edge, invisible safety rails glowing beneath its wooden limbs. · Illustration: The Agentic Times

doxx.net has raised $38 million to fund a platform intended to stop AI agents causing harm as they act on the open internet, according to a SecurityWeek report published on 3 October 2026.

The company's product is described by SecurityWeek as an ADN platform. Its stated purpose is to prevent what the report calls "agentic misadventure" while an agent is operating under the authority of its user. In other words, the problem the vendor says it is addressing is not an outsider breaking in, but a legitimate agent, holding legitimate credentials, doing something its owner did not intend.

That framing matters because it sits awkwardly with how most existing controls work. Traditional access controls ask whether a request carries valid permission. An agent delegated a user's authority passes that test by design. Any action it takes inside the limits of those permissions looks authorised, even if the user would never have approved it had they been asked. doxx.net's pitch, as reported, is to intervene at that point.

SecurityWeek's report is the only source for the funding round. The material available does not name the investors, state the round's stage, or give a valuation. It also does not set out how the ADN platform works in technical terms, what it inspects, where it sits in relation to an agent's traffic, or whether it is generally available. Customer numbers and pricing are likewise not stated.

Readers evaluating the claim should treat it as a vendor positioning statement reported by a trade outlet rather than as independently tested capability. No benchmark results, third-party assessment or published research accompanies the announcement in the source material. The size of the raise says something about investor appetite for tooling aimed at autonomous agents; it says nothing on its own about whether the approach works.

The broader category the company is entering is new enough that terminology is still unsettled, and buyers comparing products in this space will find few common definitions of what "containing" an agent means in practice. Until doxx.net publishes more detail, or customers describe deployments, the specifics of its control model remain unclear. The Agentic Times has not independently verified the funding figure.

Sources

  1. securityweek.comhttps://securityweek.com/doxx-net-raises-38-million-to-prevent-ai-agent-on-the-internet-misadventures