CVE-2026-75760: AshAi vectorize change leaks raw embedding-provider errors, including credentials, in a user-facing error

Q: Which deployments are exposed to credential disclosure?

Deployments using ash_ai versions from 0.1.0 before 1.0.0 are affected when vectorization failures can be returned through AshJsonApi or AshGraphql. Exposure is greatest where the embedding HTTP client retains outbound request details in its error structure, including an Authorization header containing the provider API key.

Q: What does an attacker need to trigger the leak?

An attacker needs a way to submit content that is vectorized and cause the embedding-provider request to fail. Oversized or malformed vectorized content can make these failures attacker-reachable.

Q: How can I determine whether an instance may already be affected?

Check whether the application uses AshAi.Changes.Vectorize on an ash_ai version earlier than 1.0.0 and exposes validation errors through AshJsonApi or AshGraphql. Review user-facing vectorization failure responses for provider request URLs, provider response bodies, or Authorization header values.

Q: What can be done before upgrading?

Avoid returning raw vectorization errors to callers and replace them with a generic error message, while logging the underlying error only in protected application logs. Restrict or validate vectorized input to reduce attacker-triggered provider failures, and rotate provider API keys if they may have been exposed.

Published Aug 31, 2026

Updated

Generation of Error Message Containing Sensitive Information vulnerability in ash-project ashai discloses provider request state and credentials in a user-facing validation error.

In AshAi.Changes.Vectorize, when the embedding provider call fails the change added a changeset error whose message inspected the raw error term (An error occurred while generating embeddings: #{inspect(error)}). A plain-string adderror produces an Ash.Error.Changes.InvalidChanges in the :invalid class, which AshJsonApi and AshGraphql render back to the caller. The embedding client's error term is not sanitized, so it can carry the request URL, the provider response body, and, for HTTP clients that keep the request in the error struct, the outbound Authorization header with the provider API key. Failures are attacker-reachable via oversized or malformed vectorized content. The fix logs the raw error and returns a generic message.

This issue affects ashai: from 0.1.0 before 1.0.0.

Affected Software

1 affected component

ash-project ash_ai<1.0.0

Remediation

Recommended actions to resolve this vulnerability, in priority order.

Upgrade
Upgrade ash_ai to a version that resolves this vulnerability.
Fixed in 1.0.0
Operational
Ensure the embedding-provider errors that may have included credentials are no longer logged to user-facing validation errors; after upgrading, review logs for any previously exposed credentials/state and treat them as potentially compromised.

Event History

Aug 31, 2026

CVE Published

via MITRE·01:09 AM

Data Sourced

via MITRE·01:09 AM

DescriptionWeakness

Frequently Asked Questions

Which deployments are exposed to credential disclosure?

Deployments using ash_ai versions from 0.1.0 before 1.0.0 are affected when vectorization failures can be returned through AshJsonApi or AshGraphql. Exposure is greatest where the embedding HTTP client retains outbound request details in its error structure, including an Authorization header containing the provider API key.

What does an attacker need to trigger the leak?

An attacker needs a way to submit content that is vectorized and cause the embedding-provider request to fail. Oversized or malformed vectorized content can make these failures attacker-reachable.

How can I determine whether an instance may already be affected?

Check whether the application uses AshAi.Changes.Vectorize on an ash_ai version earlier than 1.0.0 and exposes validation errors through AshJsonApi or AshGraphql. Review user-facing vectorization failure responses for provider request URLs, provider response bodies, or Authorization header values.

What can be done before upgrading?

Avoid returning raw vectorization errors to callers and replace them with a generic error message, while logging the underlying error only in protected application logs. Restrict or validate vectorized input to reduce attacker-triggered provider failures, and rotate provider API keys if they may have been exposed.

CVE-2026-75760: AshAi vectorize change leaks raw embedding-provider errors, including credentials, in a user-facing error

Affected Software

Remediation

Event History

Frequently Asked Questions

Which deployments are exposed to credential disclosure?

What does an attacker need to trigger the leak?

How can I determine whether an instance may already be affected?

What can be done before upgrading?

Company

Resources

Contact

Frequently Asked Questions

Which deployments are exposed to credential disclosure?

What does an attacker need to trigger the leak?

How can I determine whether an instance may already be affected?

What can be done before upgrading?