The Salesloft Drift Hack and The Need for Agent Security
By Jim Lundy
The Salesloft Drift Hack and the Need for Agent Security
The interconnected SaaS ecosystem that powers modern business is built on a network of non-human workers: chatbots, APIs, and automated integrations. The catastrophic Salesloft Drift hack has just provided a brutal lesson in what happens when the identities of these agents are left unsecured. In a move that sent shockwaves through the industry, Salesloft took its entire Drift platform offline following a massive breach, validating the urgent need for a new approach to security. This blog analyzes the incident through the lens of a critical new category Aragon Research identified in June 2025: Agentic Identity and Security.
Why Was the Breach So Devastating?
At its core, the attack on the Salesloft Drift platform was a classic case of stolen identity—not a human’s, but a machine’s. The threat actor, UNC6395, compromised OAuth tokens associated with the Drift service. These tokens are not just simple keys; they represent the trusted identity of the Drift software agent, granting it permission to act on a user’s behalf within integrated applications like Salesforce. Once the attacker possessed this trusted agentic identity, they could freely access and exfiltrate data from the hundreds of customer environments connected to Salesloft Drift, leading to one of the most significant supply chain attacks in the MarTech space.
Analysis: A Systemic Failure of Non-Human Identity Management
For years, the cybersecurity world has focused relentlessly on protecting human identities with tools like multi-factor authentication and identity access management (IAM). This was, and still is, critical. However, this focus has created a massive blind spot. The modern enterprise now runs on thousands of non-human agents that communicate and transact with each other constantly. The Drift chatbot is a perfect example of such an agent. Each of these agents has an identity and a set of permissions, yet we lack the specific frameworks and platforms to govern them.
The Salesloft Drift hack is the poster child for this systemic failure. It wasn’t a user’s password that was phished; it was the identity of an application agent that was stolen. This is precisely the scenario that led Aragon Research to define the Agentic Identity and Security market in our recent research note. This emerging category addresses the full lifecycle of discovering, managing, and securing the credentials, permissions, and activities of non-human agents. The fallout from the Drift breach proves that without a dedicated strategy for agentic security, organizations are leaving a backdoor to their most critical systems wide open. Relying on each individual SaaS vendor to perfectly secure their own integrations is no longer a viable security posture.
What Should Enterprises Do?
The immediate triage remains the same: any organization impacted by the Drift hack must revoke credentials and audit logs for malicious activity. However, the long-term strategic response must be different. This incident must serve as the catalyst for enterprises to look beyond human-centric security.
First, organizations must begin the process of discovery. You cannot protect what you cannot see. It is imperative to inventory all non-human agents operating in your environment—from SaaS integrations and API connections to RPA bots—and map their permissions. Second, this event should fundamentally change procurement and vendor risk management. Questions must now be asked not just about a vendor’s internal security, but how they manage the identities and permissions of their own software agents. Finally, enterprises must begin evaluating the emerging class of Agentic Identity and Security solutions designed to provide centralized visibility and governance over this sprawling, and largely unprotected, attack surface.
Bottom Line
The Salesloft Drift hack was not just another supply chain attack; it was a catastrophic failure of agentic security. It highlights a dangerous gap in the market that traditional IAM tools were never designed to fill. This breach makes the case powerfully and urgently for the Agentic Identity and Security category. Enterprises can no longer afford to treat the identities of their software and AI agents as a secondary concern. To prevent the next, potentially worse, breach, organizations must adopt a new security model focused on governing the non-human workers that have become indispensable to their operations.
Our Third Transform Tour for 2025 featuring our Predictions for 2026.
Join Betsy Burton, Jim Lundy and Adam Pease to hear where markets are going over the next three years.
grow a garden calculator
September 9, 2025
This analysis hits the nail on the head regarding the critical blind spot in agentic identity security. The Salesloft Drift breach was a wake-up call we cant ignore. The framework for protecting non-human agents is clearly missing, and its time enterprises woke up and started addressing this systemic risk before its too late.