I never authorized an integration that scans my secrets

A user opened a GitHub issue after discovering Home Assistant’s pwned-passwords integration scanned local secrets and queried a breach database, saying this occurred without their authorization and describing it as a severe privacy invasion. The reporter, a security professional, argued that even hashed or k-anonymity-protected checks could be abused if a malicious actor gained access to Home Assistant infrastructure, demanded an opt-in toggle or a way to disable the scanning, and asked how to stop the data from being sent to a third party. The report referenced version 2026.10.1 and linked the pwned-passwords documentation.

A project member responded that the feature is privacy-preserving (citing k-anonymity and Cloudflare’s approach used by services like 1Password), insisted passwords are not sent, and closed the issue as completed and later as won’t fix. Other users echoed the request to provide a disable option and expressed offense at the dismissive tone; maintainers pushed back, noted the documentation location and open-source nature, and locked the conversation as too heated. The core dispute remains: users want explicit opt-in control and a clear disable path, while maintainers rely on documented privacy guarantees and refuse to change the default behavior.