Data on nearly every user of blockchain analytics company Chainalysis has been compromised after what the firm calls an “agentic cyber attack.” The incident exposed names, usernames, password hashes and active tokens—leading Chainalysis to take major products offline and force widespread credential resets.
Chainalysis confirmed it discovered the breach during an ongoing investigation of a multi-front attack on its systems. In response, it disabled access to Reactor and Data Solutions. The firm stressed there’s no evidence that investigation data or attribution intelligence was viewed or tampered with. Still, the breach involved not only stored passwords but also live authentication tokens for all users.
To limit further exposure, Chainalysis deactivated its Know Your Transaction (KYT) service UI and API, along with address screening tools and the login page. The company said it now requires password resets and API key rotation once disrupted services are restored.
Affected accounts have already been locked, and customers who received password-reset notifications should note that resetting isn’t immediately possible. Chainalysis urges anyone using the same credentials elsewhere to change them straightaway, even before recovery begins.
How It Happened
The breach stemmed from what Chainalysis describes as an “on-path attack,” which revealed a subset of customer passwords. That vector has since been closed. While the firm continues its internal investigation, it is restoring access to KYT and Address Screening—but only under stricter safeguards. These include allowlisting known IPs, giving customers visibility into trusted access points, and mandating credential changeovers.
The full scale of the attack isn’t clear, and the perpetrator remains unidentified. Chainalysis noted the incident resembles recent breaches at other tech platforms, particularly in its use of agentic AI-powered tools to orchestrate widespread detection of exposed secrets.
A Broader Pattern: AI-Enhanced Breaches
Chainalysis isn’t alone. Multiple companies—including Chainstack, OpenCode, Warp, and WorkOS—have recently disclosed being targeted by AI-driven attacks involving the use of an autonomous framework known as ARTEX. These attacks scan public repositories for credentials and secrets, then rapidly escalate what they find into deeper intrusions.
Microsoft has warned that agents using ARTEX have been active since at least August 2026, detecting exposed secrets in GitHub, API keys, cloud credentials, and AI/ML infrastructure. One such agent reportedly managed to move from public repo to Azure Key Vault compromise in just five minutes. There’s also a surge in misuse of stolen access for AI-heavy operations—especially large language model (LLM) misuse—where compromised AI service credentials are sold or used to run unauthorized workloads.
The resale economy for stolen cloud-based AI access is now in full swing. Services dubbed “LLM Mixers” are offering discounted access to model APIs that weren’t theirs to sell, masking their origins, and enabling thousands to use stolen credentials without having to set any up themselves.
Some of the cloud-based services now under fire include Microsoft’s Azure, OpenAI endpoints, OAuth apps, and other enterprise AI infrastructure. Misused credentials have given attackers access to storage, Kubernetes clusters, AI model APIs, and even internal documents and mailboxes. This kind of attack vector is becoming alarmingly efficient and automated.
While Chainalysis is working to restore its platforms with tougher credential controls and enhanced monitoring, this incident highlights the vulnerabilities organizations face when using shared secrets, overloaded credentials, and legacy access models. In the age of AI-powered attacks, those weak points are now much more likely to be exploited.
What this means: Chainalysis’s breach underscores a major shift in how attacks are carried out—AI-savvy adversaries are not just snooping for weak passwords but automating intrusion across cloud services. Organizations that rely on predictable credentials, public repos, or reuse are at far greater peril. Moving forward, zero-trust credential hygiene, routine secrets audits, agent detection, and enforcing rotation policies aren’t optional—they’re essential tools in this new threat landscape.
