Chainalysis breach widens: password hashes and live tokens for all users exposed after "agentic" intrusion

Update, November 5, 2026: Chainalysis has significantly widened its account of the breach in updates to its incident page on November 2 and 4. It now says that "names, usernames, password hashes and live tokens for all Chainalysis users were exposed". It says a subset of customers' passwords was exposed in an "on-path attack". It is also contacting customers "whose KYT accounts appear to have been accessed without authorization". Several products, the login page and password resets are offline. This article has been rewritten to reflect those updates.

Chainalysis, whose blockchain-tracing software is used by law enforcement, financial institutions and compliance teams, first said on October 30, 2026 that an outside attacker using what it calls "offensive agentic techniques" got into its systems and internal tools in October. At the time it listed only user IDs, names, email addresses and role labels. Its updates since then go much further. Password hashes and live tokens for all users were exposed. Some passwords were intercepted in transit. Some accounts on its KYT transaction-screening product appear to have been accessed without authorisation. The company has taken several products offline. It says it still has no evidence that investigation data was accessed. If your organisation uses Chainalysis, plan to reset every user's password and rotate API keys as soon as the company reopens access. Change any reused passwords elsewhere now.

What happened, in plain English

Chainalysis sells software that follows the money on public cryptocurrency ledgers. Its customers have included the US Federal Bureau of Investigation (FBI), the Drug Enforcement Administration, IRS Criminal Investigation and the UK's National Crime Agency, as well as banks and brokers. In other words, its user list is a list of people who investigate crypto crime.

October 30. In a notice to customers, later published on its blog, Chainalysis said it "detected an external actor using offensive agentic techniques to gain unauthorized access to Chainalysis systems and internal tools." "Agentic" here refers to AI agents: software built on a large language model (LLM) that does not just answer questions but takes actions on its own, such as scanning, trying logins, running commands, and retrying when something fails. The company listed "user IDs, full names, email addresses, and organization role labels" as accessed. It said it had found no indication that anyone accessed investigation content ("such as graphs, notes, or cases"), transaction or query results, passwords in clear text, multi-factor authentication (MFA) secrets or session tokens. It also said the incident was "not related to the recent crypto theft at Binance."

November 2. Chainalysis took Reactor, its investigation platform, and Data Solutions offline as a precaution while outside experts ran security assessments. Its compliance transaction-monitoring products stayed online.

November 4, morning. Chainalysis emailed certain customers to say their passwords had been exposed in an "on-path attack", meaning an attacker positioned between the user and the service. The email describes an "ongoing multifront cyberattack". It says the vector "has been contained and remediated", that affected accounts were locked as a precaution, and that "we have no evidence that your account has been improperly accessed." Later that morning, citing "abnormal activity", the company took the KYT interface and API, other address-screening products and its login page offline. Password resets became unavailable.

November 4, 11:00 p.m. EST. Chainalysis said that "On Monday" (November 2) it had discovered that "certain customer information including names, usernames, password hashes and live tokens for all Chainalysis users were exposed to a malicious actor." It said it is contacting customers "whose KYT accounts appear to have been accessed without authorization". It also said it "continue[s] to have no evidence that any investigations data has been accessed", and reported no evidence that its attribution data had been manipulated. It planned to begin restoring KYT and Address Screening on November 5 for customers who register trusted IP addresses for an allowlist. Reactor, Data Solutions and Chain (Beta) remain offline with no return date. Every user will have to reset passwords and rotate API keys "as appropriate".

These later statements partly overturn the first notice. On October 30, the exposed data looked like a staff directory. By November 4 it included credential material.

Are you affected? What to do now

Who should care: anyone with a Chainalysis account, and the IT and security teams that support them. That includes investigators and compliance and anti-money-laundering staff. Chainalysis says the November 4 exposure covers all users. It has promised each organisation a breakdown of its own records. If you do not use Chainalysis, there is nothing to do today beyond the general lesson at the end of this section.

Checklist for Chainalysis customers

• Register your trusted IP addresses with Chainalysis or your account team. The company says this allowlist is the route back into KYT and Address Screening, and to password resets and API-key rotation.
• Reset every user's password and rotate API keys as soon as Chainalysis grants access. Treat the old ones as compromised, since password hashes and "live tokens" were exposed.
• Change reused passwords now. Chainalysis tells users to reset passwords "on other systems where you may be reusing credentials." Hashes can be cracked offline, and Chainalysis has not said how its passwords were hashed.
• Review KYT activity for your organisation and wait for Chainalysis's notice if your accounts were among those that "appear to have been accessed without authorization".
• Enable two-factor authentication wherever it is offered. Chainalysis "strongly" recommended it in its first notice.
• Brief users today. Names, usernames, emails and job roles are exactly what a convincing targeted phishing email needs. Tell users to expect lures that mention Chainalysis, the incident, password resets or their investigative role, and to reach the service directly rather than through emailed links. Chainalysis tells users to be "hyper-vigilant" about unexpected reset requests.
• Keep the per-organisation breakdown Chainalysis has promised. Your privacy or legal team may need it to assess notification duties under your local rules.
• Ask your account contact what was exploited, what "live tokens" means for your integration, how long the attacker had access, and whether the on-path attack, the token exposure and the KYT access were the same campaign. The public updates do not say.

For everyone else: a supplier breach that starts as "non-sensitive" contact data can grow. Check which of your vendors hold your staff's credentials and API keys, and make sure you can rotate them quickly.

Chainalysis has not published indicators of compromise (IOCs), and we found no CISA or other government advisory about this incident.

The expert view

What is new, and what is not. This is one of the first breach notices we have seen in which a company states, without hedging, that the intruder used AI agents. Usually that claim comes later from a threat-intelligence firm, or not at all. The second notable thing is how far the scope moved in five days. "No clear-text passwords" (October 30) and "password hashes and live tokens for all Chainalysis users" (November 4) can both be true. But they read very differently to a customer deciding whether to act.

What "offensive agentic techniques" probably means, conceptually. Chainalysis does not define the phrase, and its November updates do not mention AI at all. Across public reporting this autumn, attacker agents follow a familiar pattern. A human sets a goal and a plan. An LLM-driven agent then carries out reconnaissance, tries known weaknesses, keeps going after errors, and moves quickly once inside. The steps themselves are textbook. The change is speed, persistence and the low skill needed to run them. Chainalysis's mention of patching "the exploited vulnerabilities" fits that picture of known weaknesses found and used quickly. We do not know whether a commercial model, an open-weight one or a criminal toolkit was involved. Chainalysis declined to answer The Block's follow-up questions, and a spokesperson told The Record it could not share more than the blog.

Why the data matters. For a firm whose users investigate criminals, the user list is itself intelligence: who works crypto cases, at which agency or bank, in which role. Password hashes and live tokens turn that list into a possible way in. They allow offline cracking, reuse against other services, and possibly direct access to customer integrations. The most useful defences are fast credential rotation, phishing-resistant two-factor authentication and clear warnings to the people on that list.

The defender side. Chainalysis says it is using "sophisticated large language models to support a broader review of our platform's security", alongside outside incident-response experts. This mirrors a wider shift: the same class of tools is now used on both sides, and the speed of the defensive review matters as much as the speed of the attack.

The broader trend. This fits a run of reports this autumn. CrowdStrike, for example, described four separate criminal crews using AI coding agents to run intrusions and assessed that agentic AI is "commoditizing" basic offensive operations. Chainalysis's claim is consistent with that trend, but it has published no technical evidence, so the AI attribution rests on the company's word for now.

Still unknown:

• How the attacker got in, which vulnerabilities were exploited, and which internal tools were reached.
• What the "on-path attack" was, and whether it, the hash-and-token exposure and the KYT account access involve the same attacker or methods.
• What "live tokens" are, and how passwords were hashed.
• How long the access lasted, and how many customers and users are affected. Chainalysis now says "all" users were exposed but gives no number.
• What evidence led Chainalysis to call the techniques "agentic".
• Who the attacker is. Chainalysis names no actor, and nobody has made a public attribution.
• When Reactor, Data Solutions and Chain (Beta) will return.

Official sources

• Chainalysis, "Chainalysis Security Incident Information", October 30, 2026, updated November 2 and November 4 (latest update 11:00 p.m. EST).
• CrowdStrike, "CrowdStrike Uncovers New Agentic AI Campaigns", October 21, 2026 (context on the wider trend).
• Coverage: The Block, "Chainalysis says hacker using 'offensive agentic techniques' accessed its customers' names and emails", October 31, 2026.
• Coverage: The Hacker News, "Chainalysis Confirms Intrusion Into Internal Tools and Exposure of Customer Identity Data", October 31, 2026.
• Coverage: The Record, "Chainalysis warns all users hit in cyber breach involving 'agentic' techniques", November 5, 2026.