HighBreach / campaignMonitoring

Botpress breach: auth proxy flaw exposed bot configs in 16 workspaces

Botpress says an intruder read bot configurations in 16 workspaces; if you were contacted, treat every stored credential as exposed and rotate it.

Published
Updated
Exploitation
Yes, Botpress states an unauthorized party used an authentication proxy flaw and a staff API credential, reading bots in 16 workspaces on October 21, 2026

What happened

Botpress’s CTO says that on October 12, 2026, someone began probing the authentication proxy in front of several internal tools. The proxy had a flaw in how it validated sign-ins, and the attacker used it to enter those tools with sessions that appeared to belong to staff (Botpress). Botpress found no evidence that any employee’s identity-provider account was compromised (Botpress).

Between October 13 and October 21, the attacker had access to internal tools and some connected systems. That access gave visibility into customer account information and internal data, including data related to staff API credentials (Botpress).

On October 21, between 07:44 and 08:23 UTC, the attacker used one employee’s API credential and its admin access to read bot configurations in 16 customer workspaces. The employee had admin access because they were a member of those workspaces (Botpress). A bot’s configuration includes integration settings, which can hold AI provider API keys, messaging channel tokens, CRM credentials and similar secrets (Botpress). Separately, the attacker used weaknesses in the setup flow of some integrations to change the settings of one integration in each of two other workspaces (Botpress).

Botpress fixed the proxy flaw at about 16:25 UTC on October 21, rotated its secret and revoked the employee credential. The attacker kept trying to get back in through other routes and new platform accounts; Botpress says it closed every route it identified and that attempts observed through November 3 were rejected (Botpress).

Botpress says the evidence points to one goal: obtaining AI usage to resell on AI black markets. It reports no ransom demand and no sign that customer data was published or offered for sale (Botpress). Unite.AI reports the same account and the same attribution to Botpress’s disclosure (Unite.AI).

What Botpress says it has not found: evidence of changes to source code, build pipelines or published packages, or of access to end-user conversations handled by bots. It adds that it does not store full payment card numbers (Botpress).

Not confirmed: Botpress’s logs cannot show exactly which of the read bots held a valid credential at the time (Botpress). No CVE identifier is given in the sources.

Am I affected?

  • You were contacted by Botpress: you are in scope. Botpress says it contacted the owners of the 16 workspaces where bots were read and of the two where integration settings were changed, and sent each the list of bots and integrations involved (Botpress).
  • You were not contacted: Botpress says it has no evidence your bots or their configuration were accessed or changed (Botpress).

The sources do not say which versions or plans are involved. The flaw was in Botpress’s internal tooling, not in a package customers install.

What to do now

  1. Find the message from Botpress and the list of bots and integrations it names.
  2. Treat every credential stored in those bots as exposed, as Botpress asks, because it cannot tell which were valid at the time (Botpress).
  3. Rotate those credentials at the third-party services themselves: AI provider API keys, messaging channel tokens, CRM credentials and similar (Botpress).
  4. If you are one of the two customers whose integration settings were changed, review the settings of the integration named in your notice. The sources do not describe what was altered.
  5. Botpress says it will work directly with affected customers until their credentials are rotated, so reply to its contact if you need help (Botpress).
  6. Watch the Botpress disclosure for updates; the company says it will revise the post if its findings change (Botpress).

Botpress says remediation is still under way: internal credential rotation, a review of internal tool access and staff permissions (including when employees should hold admin access to customer workspaces), and a change to the bot API so stored bot credentials cannot be reached this way again (Unite.AI).

Update, 17 November 2026

  • Botpress says that before and during the intrusion the attacker ran a large network of accounts on its platform to consume AI credits, and much of what they did inside Botpress systems was aimed at keeping those accounts running without limits. (botpress.com)
  • Botpress says it blocked the accounts, workspaces and network routes associated with the attacker, hardened its ingress rules, and restricted and restarted the affected internal tools. (botpress.com)
  • Botpress says it reversed the changes the attacker made to internal settings, strengthened how integration configuration is stored, and started rotating the internal credentials that appeared in the files the attacker read. (botpress.com)

Sources