Tune Chrome Extension Limits Toxic Web Comments

Jigsaw, an Alphabet incubator working on difficult global security challenges, recently announced a Chrome extension that allows users to limit the toxicity seen in comments while browsing. The extension, Tune, is currently an experimental extension. During this experimental phase, Tune works on many popular platforms including YouTube, Facebook, Twitter, Reddit, and Disqus.

Tune is powered by the Perspective API which Jigsaw released in 2017. The Perspective API uses Machine Learning to detect abusive and toxic speech. The AI was originally trained through human toxicity scores given to hundreds of thousands of internet comments. The more the Perspective API is used, the more accurate it becomes.

Tune, nor Jigsaw, saves comments. The text is sent to the Perspective API for scoring and is then deleted. Currently, Tune is limited to English language comments. Tune is opened source and available at GitHub. Tune is one tool in the broader Conversation AI research project which is also available at GitHub.

Be sure to read the next Security article: Google Chromium Flaw Put Android Users' Data at Risk