❎ moar anti-AI measures

README.md

@@ -26,7 +26,7 @@ A Flask-based webcomic website with server-side rendering using Jinja2 templates
   - [Common SEO Questions](#common-seo-questions)
 - [Content Protection & AI Scraping Prevention](#content-protection--ai-scraping-prevention)
   - [Protection Features](#protection-features)
-  - [Optional: Additional Protection Measures](#optional-additional-protection-measures)
+  - [Advanced: Image-Level Protection Tools](#advanced-image-level-protection-tools)
   - [Important Limitations](#important-limitations)
   - [Customizing Your Terms](#customizing-your-terms)
   - [Testing Your Protection](#testing-your-protection)
@@ -509,34 +509,59 @@ The Terms page is automatically linked in your footer and includes:
 - TDM rights reservation (EU Directive 2019/790 Article 4)
 - Clear permitted use guidelines
 
-### Optional: Additional Protection Measures
+#### HTTP Headers
+Sunday Comics automatically adds `X-Robots-Tag: noai, noimageai` headers to all responses for additional AI blocking enforcement.
 
-#### HTTP Headers (Advanced)
+#### TDM Reservation File
-For stronger enforcement, you can add HTTP headers. Add this to `app.py` after the imports:
+The `/tdmrep.json` endpoint formally reserves Text and Data Mining rights under EU Directive 2019/790, pointing to your Terms of Service.
 
-```python
+### Advanced: Image-Level Protection Tools
-@app.after_request
-def add_ai_blocking_headers(response):
-    """Add headers to discourage AI scraping"""
-    response.headers['X-Robots-Tag'] = 'noai, noimageai'
-    return response
-```
 
-#### TDM Reservation File (Advanced)
+For artists who want to protect their work at the image level, consider these specialized tools:
-Create a `/tdmrep.json` endpoint to formally reserve Text and Data Mining rights:
 
-```python
+#### Glaze (Style Protection)
-@app.route('/tdmrep.json')
+**What it does:** Adds imperceptible changes to images that prevent AI models from accurately learning your artistic style.
-def tdm_reservation():
+
-    """TDM (Text and Data Mining) reservation"""
+**Best for:**
-    from flask import jsonify
+- Protecting your unique art style from being copied by AI
-    return jsonify({
+- Making AI-generated imitations look wrong or distorted
-        "tdm": {
+- Artists concerned about style mimicry (e.g., "draw like [artist name]" prompts)
-            "reservation": 1,
+
-            "policy": f"{SITE_URL}/terms"
+**How to use:**
-        }
+1. Download from [glaze.cs.uchicago.edu](https://glaze.cs.uchicago.edu)
-    })
+2. Process your comic images before uploading to your site
-```
+3. The changes are invisible to humans but confuse AI models
+
+**Trade-offs:**
+- Processing time: Can take several minutes per image
+- Slight file size increase
+- Requires reprocessing all comics
+
+#### Nightshade (Data Poisoning)
+**What it does:** Makes images appear as something completely different to AI models while looking normal to humans.
+
+**Best for:**
+- Active defense against unauthorized AI training
+- Making scraped data actively harmful to AI models
+- Artists who want to fight back against scraping
+
+**How to use:**
+1. Download from [nightshade.cs.uchicago.edu](https://nightshade.cs.uchicago.edu)
+2. Process images before uploading (can combine with Glaze)
+3. AI models trained on these images will produce incorrect results
+
+**Trade-offs:**
+- More aggressive than Glaze (may violate some ToS)
+- Processing time similar to Glaze
+- Ongoing research tool, effectiveness may vary
+
+#### Recommendations
+- **Use Glaze if:** You want passive protection for your art style
+- **Use Nightshade if:** You want active defense and accept the risks
+- **Use both if:** Maximum protection is your priority
+- **Combine with Sunday Comics protections:** These tools complement the web-based protections (robots.txt, meta tags, etc.)
+
+**Note:** Both tools are free, open-source projects from the University of Chicago's SAND Lab, specifically designed to help artists protect their work from AI exploitation.
 
 ### Important Limitations
 

app.py

@@ -19,6 +19,13 @@ app = Flask(__name__)
 app.config['SECRET_KEY'] = os.environ.get('SECRET_KEY', 'your-secret-key')
 
 
+@app.after_request
+def add_ai_blocking_headers(response):
+    """Add headers to discourage AI scraping"""
+    response.headers['X-Robots-Tag'] = 'noai, noimageai'
+    return response
+
+
 @app.context_processor
 def inject_global_settings():
     """Make global settings available to all templates"""
@@ -321,6 +328,17 @@ Disallow: /
     return Response(robots_txt, mimetype='text/plain')
 
 
+@app.route('/tdmrep.json')
+def tdm_reservation():
+    """TDM (Text and Data Mining) reservation - signals AI training prohibition"""
+    return jsonify({
+        "tdm": {
+            "reservation": 1,
+            "policy": f"{SITE_URL}/terms"
+        }
+    })
+
+
 @app.errorhandler(404)
 def page_not_found(e):
     """404 error handler"""