video-analysis

MEDIUM W011: Third-party content exposure detected (indirect prompt injection risk).

• Third-party content exposure detected (medium risk: 0.65). Outsider free text can enter the LLM context via extraction mode: _transcribe_audio() sends Whisper/Gemini transcription text (derived from the video’s audio track) to the chat model, and the video content itself is not authored by the operating user (e.g., an uploaded/downloaded outsider video), so its spoken text can be treated as outsider-authored free text.

MEDIUM W021: Hidden or invisible Unicode characters detected (potential obfuscation or prompt injection).

• Hidden Unicode characters detected (1 type(s) found)