Câu đố về Deepfake

Câu trả lời và giải thích xuất hiện khi bạn làm bài kiểm tra.

1. Một tin nhắn thoại yêu cầu thay đổi tài khoản nhạy cảm. Người nhận nên xác minh nó như thế nào?
   • Liên hệ với người đó thông qua một kênh đáng tin cậy đã được ghi lại
   • Sử dụng chi tiết liên hệ có trong tin nhắn đáng ngờ
   • Hãy coi sự giống nhau về giọng nói là bằng chứng đầy đủ
   • Yêu cầu người gửi lặp lại yêu cầu
2. A detector labels a clip as synthetic. What does its score establish?
   • Proof that a specific person created it
   • A signal that needs context, not a definitive verdict
   • Proof that the event depicted never happened
   • A guarantee that the media is authentic
3. Why might a convincing image still be misleading?
   • Only audio can be manipulated
   • Visual realism proves provenance
   • Synthetic or manipulated media can depict events that did not occur
   • A content credential verifies every scene claim
4. What factors matter when deciding whether media manipulation is appropriate?
   • Whether the edit is difficult to detect
   • Only technical quality
   • The file’s resolution
   • Consent, disclosure, purpose, and rights of depicted people
5. What do provenance records or content credentials prove by themselves?
   • They prove that all depicted events occurred
   • They can show recorded history but do not prove every claim in the scene
   • They identify every editor with certainty
   • They establish consent for every use
6. A reporter investigates a suspicious clip. What should they preserve?
   • A detector score without the file
   • Only a compressed social-media preview
   • Original media and provenance information
   • An edited copy that removes metadata
7. When should synthetic media be labeled on publication?
   • When viewers might otherwise be misled
   • Only when the file is low resolution
   • Never if a detector did not flag it
   • Only when the subject complains
8. A recording is genuine, but a short crop removes the preceding question and changes the apparent meaning. What does this illustrate?
   • Only synthetic media can mislead
   • A genuine file proves every claim made about it
   • A detector can reconstruct all missing context
   • Authentic media can still mislead through cropping or relabeling