HighSecrets & credentialsBLWK-AI-001

A secret is exposed in AI assistant context

Depends on the kind of secret found.

What this means

A provider secret (Anthropic, OpenAI, GitHub, AWS, …) in an assistant's context, config, or transcript.

How to fix it

Remove the key from the file and replace it with a new one. Anything pasted into an AI chat should be treated as leaked.

1. Use “Remove secret from this file” below. Bulwark backs up the original first.
2. Create a new key with the provider and delete the old one.
3. Next time, keep keys in a password manager (Bitwarden, 1Password, KeePassXC) and pass them to tools as environment variables, for example with op run or bws run, instead of pasting them into the chat.

Bulwark can do this for you

In the app, open the issue and use “Remove secret from this file”. From a terminal: bulwark ai redact (preview), then bulwark ai redact --apply.

Check it yourself

Run a scan and look for BLWK-AI-001 in the results:

bulwark scan
bulwark ai scan

Installed the standalone CLI package instead of the app? The command is bulwarkctl.

References

• MITRE ATT&CK T1552.001

Related checks

• A long-lived AWS access key is stored on disk · BLWK-CRED-001
• Git stores passwords and tokens in plain text · BLWK-CRED-002
• Docker registry logins are stored in the config file · BLWK-CRED-003
• A credential file can be read by other users · BLWK-CRED-004
• SSH private key is not passphrase-protected · BLWK-SSH-012
• SSH private key passphrase status could not be determined · BLWK-SSH-013