NOAIS: A Browser Extension to Combat AI Content Overload, Bringing Web Pages Back to Authenticity

NOAIS is an open-source browser extension, whose full name is "No Artificial Intelligence Slop". Its core mission is to detect and block AI-generated content (AI slop) on web pages, helping users filter low-quality text and deepfakes. Its features include:

• 100% local operation to protect user privacy (no network requests, zero dependencies)
• Supports Chrome and Firefox browsers
• Covers mainstream platforms like YouTube and Facebook

The project is maintained by nedaktov-ops, with its source on GitHub. Release date: 2026-06-07. Original link: https://github.com/nedaktov-ops/NOAIS

Core point: NOAIS provides a privacy-first AI content filtering tool through local heuristic algorithms, giving users back the initiative in content selection.

In recent years, generative AI technology has exploded. While it brings convenience to content creation, it also gives rise to the "AI Slop" (AI garbage content) problem. A large number of low-quality AI-generated texts, images, and videos drown out real human creations. These contents are often created to manipulate algorithms, promote scams, or drive traffic, seriously damaging users' web experience. Against this background, NOAIS came into being.

NOAIS is a completely free and open-source browser extension. Its mission is to help users detect, block, and disable AI-generated content on web pages, covering mainstream platforms such as YouTube, Facebook, Instagram, TikTok, Twitter/X, Reddit, and LinkedIn.

Unlike solutions that rely on cloud services, NOAIS insists on 100% local operation: all detection algorithms run on the user's device, with zero network requests, zero dependencies, and no need to download large AI models. This ensures that users' browsing data is not uploaded, maximizing privacy protection.

NOAIS v0.3.0 uses heuristic text analysis methods to measure the "human characteristics" of text through statistical indicators. The four core indicators are as follows:

1. Burstiness: Calculates the coefficient of variation of sentence length. Human writing has an irregular rhythm, while AI text has a uniform rhythm.
2. Type-Token Ratio (TTR): The ratio of unique words to total words. Humans use richer vocabulary, while AI tends to repeat words.
3. Shannon Entropy: Measures the uniformity of word distribution. Human writing has higher entropy (more jumps in thinking).
4. Hapax Legomenon Ratio: The proportion of words that appear only once in the text. Human writing has more such words.

A comprehensive score of AI possibility from 0 to 100 is output: human text scores about 23 (green), AI text scores about 81 (red), providing a basis for filtering decisions.

NOAIS adopts a progressive iteration strategy:

• v0.1.0: Built the core framework, supporting enable/disable toggle, state persistence, dark mode, and accessibility design.
• v0.2.0: Introduced a phrase counter that scans 5 AI signature phrases (e.g., "As an AI language model").
• v0.3.0: Implemented a multi-indicator scoring system, combining phrase counting and stylometric analysis. Users can view the AI possibility score bar and the number of detected phrases.

NOAIS's future version plans:

• v0.4: Introduce per-site independent settings and a sensitivity adjustment slider.
• v0.5-v0.6: Develop adapters for YouTube (video comments, descriptions) and Facebook (social posts, comments).
• v0.7: Consider introducing an optional small ONNX model (<5MB) to improve detection accuracy.
• v1.0: Officially release on GitHub Releases and Firefox Add-ons.

Currently, NOAIS is in the active development phase (v1.0 not yet released). Early experience method: Clone the repository and load the unpacked extension (directly load the extension/ directory, zero dependencies and no build required).

The project uses the Apache License 2.0 open-source agreement. Community contributions are welcome: Participate in improvements via GitHub Issues or Pull Requests.

NOAIS does not deny the value of AI technology, but provides users with the right to choose in the era of AI content overload. Under information overload, independently deciding which content to consume is an important digital literacy. NOAIS uses an open-source, local, and privacy-first design to provide a technical solution for the discussion of content authenticity, reminding us that technological progress should serve human choice rather than deprive it.