feat: add screenshot-specific disclosure and court brief conclusion text

- Forensic report: adds screenshot disclosure explaining that photographic forensic
tests don't apply to UI screenshots, LCD pixel grid artifacts mimic diffusion
harmonics, and verdict should be interpreted with this limitation in mind
- Court brief: overrides conclusion/recommendation text when screenshot detected,
recommending manual content review instead of photographic forensics

explanation.py

@@ -179,6 +179,28 @@ are normal for images shared via WhatsApp, Telegram, Signal, etc. The following
 been suppressed: EXIF Completeness, Compression Ghosts, ICC Profile, Maker Note,
 Thumbnail Check, Software Detection, JPEG Quantization, and CFA Nyquist.
 
+"""
+        elif detected_modality == "SCREENSHOT":
+            report += """**Screenshot / screen capture disclosure:** This image appears to be a screen capture
+or photograph of a digital display. **FORENSIQ's methodology is designed for photographic
+images captured by cameras, not for UI screenshots or screen recordings.**
+
+Screenshots inherently lack camera sensor characteristics (no Bayer CFA pattern, no PRNU
+fingerprint, no lens physics) and have color distributions that violate natural-image
+statistical priors (few exact UI colors, dark theme backgrounds, Benford's Law violations
+in DCT coefficients). If the screenshot was captured by *photographing a screen*, the
+display's LCD/OLED pixel grid creates periodic frequency artifacts that mimic diffusion
+model noise harmonics — these are display physics, not AI generation signatures.
+
+The following test categories have been suppressed or heavily discounted: all optical lens
+tests, all sensor characteristic tests, frequency-domain diffusion detection, and
+natural-image statistical priors (Benford's Law, Color Histogram, DCT/Wavelet Kurtosis).
+
+**Interpretation guidance:** For screenshots, FORENSIQ can assess whether visible text is
+coherent and whether semantic content is plausible, but cannot make reliable determinations
+about image authenticity using signal-processing methods. The verdict should be interpreted
+with this limitation in mind.
+
 """
 
     report += f"""---
@@ -263,7 +285,24 @@ def generate_court_brief(verdict: ForensicVerdict) -> str:
     timestamp = datetime.datetime.now().strftime("%B %d, %Y at %H:%M UTC")
     
     # Determine language based on verdict
-    if verdict.verdict in ["FAKE", "LIKELY FAKE"]:
+    # Check if this is a screenshot — override conclusion text if so
+    modality_info = verdict.reasoning_tree.get("modality", {})
+    is_screenshot = modality_info.get("detected") == "SCREENSHOT"
+    
+    if is_screenshot:
+        conclusion = ("the submitted media appears to be a screen capture or photograph of a digital "
+                      "display interface, not a photographic image captured by a camera. FORENSIQ's "
+                      "signal-processing methodology is designed for photographic images and has "
+                      "limited applicability to screenshots. The visible content (text, UI elements) "
+                      "appears consistent with a genuine application interface, but this assessment "
+                      "is based on semantic and typographic analysis rather than physical forensics")
+        recommendation = ("This image is a screen capture, not a photograph. Standard photographic "
+                         "forensic tests (lens physics, sensor noise, frequency analysis) are not "
+                         "applicable to this image type. To verify the authenticity of the depicted "
+                         "content, manual review of the application data, cross-referencing with "
+                         "the original source system, or metadata analysis of the screen capture "
+                         "file itself is recommended.")
+    elif verdict.verdict in ["FAKE", "LIKELY FAKE"]:
         conclusion = "the analyzed media exhibits multiple physical and statistical anomalies inconsistent with authentic photographic capture"
         recommendation = "This evidence supports the conclusion that the media has been synthetically generated or significantly manipulated."
     elif verdict.verdict == "INCONCLUSIVE":