I built a Chrome extension that analyzes per-article news bias in the browser — no account, no server, BYOK for the deeper analysis

I'm the developer. Free, open-source (AGPL-3.0), no telemetry on what you read.

What it does. Click the toolbar icon on any news article and a side panel opens with a readout of how the article is slanted: overall lean, loaded words, headline-vs-body drift, and the specific phrases the model flagged — which get underlined inline on the page so you can see the evidence in context.

Two layers.

In-browser, no key: uses the BABE lexicon (2,100+ peer-reviewed loaded terms), attribution-verb skew, headline-vs-body drift, hedge-word count. Zero network calls.

With your own LLM key (STRONGLY RECOMMENDED) : four-dimensional breakdown (word choice, framing, headline accuracy, source mix). Every flagged phrase has to cite text from the article. Goes straight from your browser to Anthropic / OpenAI / Google.

Privacy posture. No backend exists to receive data.…