Docs

AI Shield

Detect AI-generated content in your writing and humanize it to sound natural before submission

AI Shield

AI Shield is UniFlow's built-in AI content detection and humanization tool. It analyzes your writing to estimate how likely it is to be flagged by AI detectors, then offers one-click humanization to make your text sound more natural — all without leaving the app.

Plan requirement: AI Shield is available on Pro and Max plans. Free and Starter users get one free detection to try the feature.

How It Works

AI Shield combines two industry-leading services:

  1. Detection — Powered by Phrasly's AI detection engine, which scores your text on a 0–100% scale for AI probability
  2. Humanization — Powered by Undetectable.ai's v11 engine, which rewrites text to sound naturally human while preserving meaning

Using AI Shield

Step 1: Input Your Text

Open AI Shield from the sidebar and input your text using one of these methods:

  • Paste text — Paste directly from your clipboard
  • Type — Write or edit text in the built-in editor
  • Import from clipboard — Click the clipboard button to auto-paste
  • Pick a document — Select from your saved documents in UniFlow

Step 2: Analyze

Click Analyze to run AI detection. The system will:

  1. Send your text to the detection engine
  2. Calculate an overall AI probability score (0–100%)
  3. Highlight individual sentences by their AI likelihood
  4. Show a multi-detector breakdown

Step 3: Review Results

After analysis, you'll see:

Overall Score

A large percentage indicator showing the estimated AI probability:

  • 0–30% (Green) — Low risk. Text appears human-written
  • 30–60% (Yellow) — Medium risk. Some patterns may trigger detectors
  • 60–100% (Red) — High risk. Text is likely to be flagged as AI-generated

Sentence-Level Analysis

Each sentence in your text is color-coded:

  • Green — This sentence reads as human-written
  • Yellow — This sentence has some AI-like patterns
  • Red — This sentence is likely to be flagged

Click on any sentence to see its individual score and specific issues.

Multi-Detector Panel

AI Shield checks your text against multiple detection algorithms and shows results in a radar chart, giving you a comprehensive view of how different detectors evaluate your writing.

Highlighted Text Viewer

Toggle between the original text and a highlighted view that visually marks AI-detected passages, making it easy to identify which sections need the most attention.

Step 4: Humanize

If your score is higher than desired, click Humanize to rewrite the text:

  1. Configure humanization settings (optional — see below)
  2. Click Humanize
  3. Wait for the AI to process (typically 10–30 seconds)
  4. Review the humanized version side by side with the original
  5. Copy the result or run detection again to verify improvement

Humanization Settings

Customize how aggressively the text is rewritten:

Strength

Level Description
Light (Quality) Minimal changes. Preserves most of the original phrasing
Medium (Balanced) Moderate rewriting. Good balance between naturalness and meaning
Heavy (More Human) Aggressive rewriting. Maximum humanization, may alter phrasing significantly

Style

Style Best For
Academic (Essay) Research papers, essays, formal writing
Casual (General Writing) Blog posts, emails, informal content
Business (Business Material) Reports, proposals, professional documents
Creative (Story) Creative writing, narratives

Readability Level

Level Description
High School Simpler vocabulary and sentence structure
University Standard academic reading level (recommended)

Viewing Changes

After humanization, you can:

  • Toggle Show Changes — Highlight the differences between original and humanized text
  • Compare side by side — See original and humanized versions together
  • Copy — Copy the humanized text to your clipboard
  • Re-detect — Run detection again on the humanized version to verify the improvement

History

AI Shield maintains a history of your analysis sessions:

  • View past analyses with their scores and timestamps
  • Re-open previous sessions to review results
  • Track your improvement over time

History is stored locally in your browser.

Workflow Integration

AI Shield fits naturally into UniFlow's writing workflow:

  1. Draft with UniMind or the Document Editor
  2. Polish with the Writing Assistant
  3. Check with AI Shield
  4. Humanize if needed
  5. Verify by running detection one more time
  6. Export your final document

Important Notes

Academic Integrity

AI Shield is designed to help you produce natural-sounding writing. It is your responsibility to:

  • Follow your school's academic integrity policies
  • Ensure your submissions represent your own understanding
  • Use AI tools responsibly and ethically
  • Disclose AI tool usage when required by your institution

Accuracy

No AI detection tool is 100% accurate. AI detection technology is evolving rapidly, and results should be treated as estimates rather than definitive verdicts. A low score does not guarantee your text won't be flagged, and a high score doesn't necessarily mean the text was AI-generated.

Privacy

  • Your text is processed by third-party APIs (Phrasly and Undetectable.ai) for detection and humanization
  • Text is not stored permanently by these services
  • All communication is encrypted in transit
  • See our Privacy Policy for complete details

Tips

  • Check before every submission — Make it a habit to run AI Shield on any assignment that used AI assistance
  • Use humanization iteratively — Sometimes running humanization twice produces better results
  • Focus on red sentences — You can often get a passing score by manually rewriting just the highest-risk sentences
  • Adjust your writing style — Over time, reviewing AI Shield's feedback helps you naturally write in a less detectable way
  • Use the University readability level — It produces the most natural-sounding academic writing

What's Next