1. What was the approximate total value transferred in the February 2024 Hong Kong deepfake video conference fraud?
Correct. HK$200 million — approximately US$25.6 million — was transferred across five bank accounts after a deepfake video call featuring synthetic versions of multiple company executives.
The amount was US$25.6 million (HK$200 million), distributed across five bank accounts. Every participant except the targeted employee appeared to be a real executive but was in fact a deepfake.
2. What does C2PA provide that artifact-based classifiers cannot?
Correct. C2PA is a provenance standard — it provides verifiable chain of custody rather than artifact probability estimates. Fundamentally different from classifier-based detection.
C2PA provides cryptographic provenance — a verifiable record of capture and processing history. This is categorically different from classifiers that estimate how synthetic content looks.
3. What was the Content Authenticity Initiative's founding membership, and when was it established?
Correct. CAI was founded in 2019 by Adobe, Twitter, and The New York Times — an unusual coalition spanning a creative software company, a social platform, and a major news organization. It has since grown to 2,000+ members.
The founding members were Adobe (creative tools), Twitter (distribution platform), and The New York Times (journalism) — founded in 2019. The coalition has since grown to over 2,000 member organizations.
4. The Slovak election deepfake audio case demonstrates which enforcement gap specific to political deepfakes?
Correct. AFP fact-checkers identified the audio as fabricated, but Meta did not remove it before the election. The stated reason was that fact-checker input arrived after the silence period had begun and platform review was not completed before polling. The deepfake was deployed strategically within the silence period specifically to exploit this gap.
The gap was timing: platform review pipelines were not fast enough to act on fact-checker input within the narrow window of a pre-election media silence period. The deepfake was deployed strategically two days before the vote, within the silence period, and platforms did not remove it before polling closed.
5. What is the primary reason deepfake artifacts cluster at the hairline and jaw edge rather than the center of the face?
Correct. The seam where generated content meets original footage receives less training signal and accumulates the most visible errors.
Boundary artifacts occur because the seam between synthesized and real regions is where the generative model's approximation fails most noticeably.
6. What biological signal does Intel's FakeCatcher use to detect synthetic video?
Correct. Remote photoplethysmography detects blood flow patterns — synthetic faces lack genuine circulation, making rPPG signals absent or inconsistent.
FakeCatcher uses rPPG (remote photoplethysmography) — detecting blood flow via subtle skin color changes. Synthetic faces have no real blood flow to detect.
7. What is the "consent cascade" requirement?
Correct. Consent cascade addresses the technical reality that synthetic media travels widely: if a subject revokes consent, that revocation must reach not just the original creator but every downstream system that has used or trained on the content.
Consent cascade refers to the propagation of revocation downstream — the requirement that when a subject withdraws consent, that withdrawal must reach all systems (platforms, aggregators, derivative models) that already have the content, not just the original creator.
8. What does the Partnership on AI's Responsible Practices for Synthetic Media identify as the difference between "authentication" and "disclosure" as obligations?
Correct. Authentication (provenance — who/what created this) and disclosure (audience-facing notification) are distinct mechanisms addressing different information needs: technical traceability versus human understanding.
Authentication and disclosure are complementary but distinct: authentication provides technical provenance (traceable to creation source), while disclosure communicates AI-generated status to human audiences through visible, persistent labeling.
9. The UK Online Safety Act 2023 addressed synthetic NCII by:
Correct. The UK Online Safety Act 2023 criminalized the sharing of non-consensual intimate deepfakes — one of the first national laws to do so explicitly.
The Online Safety Act 2023 made sharing non-consensual intimate deepfakes a criminal offense. This was distinct from the proposed U.S. DEFIANCE Act, which focused on a civil cause of action.
10. Microsoft's VALL-E (2023) demonstrated voice cloning from a minimum of how much source audio?
Correct. VALL-E's key advance was requiring only three seconds of audio — a dramatic reduction from earlier systems that needed minutes of clean speech.
VALL-E could clone a voice from just three seconds of audio — dramatically lowering the practical barrier for voice fraud.
11. The 2022 iScience study found that human detection of StyleGAN2 faces averaged which accuracy rate?
Correct. 48.2% — slightly below random chance. The counterintuitive mechanism: StyleGAN2 faces tended toward slight symmetry and conventional attractiveness, which humans associate with trustworthiness, leading them to judge synthetic faces as more likely to be real. Short training improved accuracy to approximately 59%, still far below reliable detection.
The accuracy was 48.2% — slightly below random chance. Synthetic faces scored higher on perceived trustworthiness, causing participants to actively misidentify them as real more often than expected. This undermines any reliance on human intuition as a deepfake detection mechanism.
12. In the 2019 UK energy company voice fraud case, how much was transferred and to whom?
Correct. €220,000 was transferred to a Hungarian supplier after a call using an AI-cloned voice of the German parent company's CEO. The money was never recovered.
€220,000 was transferred to a Hungarian supplier. The CEO of the UK energy company believed he was speaking to his German parent company's CEO — a voice later analyzed as likely AI-generated.
13. According to Sensity AI's 2019 research, what proportion of deepfake targets were women?
Correct. 99% of deepfake pornography targets were women, establishing that synthetic NCII is not a generalized harm but a gender-targeted form of abuse.
The figure was 99%. Sensity AI's research established that deepfake pornography was overwhelmingly targeting women — making it a gendered harm, not a generalized one.
14. What three components make consent ethically adequate for synthetic media use of a person's likeness?
Correct. Informed (subject understands the use), specific (consent covers only the described use-case), and revocable (subject retains the right to withdraw).
The three ethical components are informed (understanding of use), specific (limited to the described application), and revocable (ongoing right to withdraw). Legal form alone does not satisfy these requirements.
15. Who or what was "deepfakes" on Reddit, and what did they release in late 2017?
Correct. The Reddit user "deepfakes" posted videos superimposing celebrity faces onto pornographic content, then released the code publicly — triggering the automated proliferation of synthetic NCII.
It was a Reddit user named "deepfakes" who posted celebrity face-swap pornographic videos and released open-source code, rapidly automating synthetic NCII creation for anyone with a photograph.
16. The Zelensky surrender deepfake of March 2022 was distributed how — making it particularly difficult to counter?
Correct. The video appeared on a hacked Ukrainian news site — lending it the credibility of a trusted local source. Zelensky responded with a genuine video from Kyiv. Meta, YouTube, and Twitter removed it, but the hacked-news-site distribution made it especially credible before removal.
The video appeared on a hacked Ukrainian news website — making it appear to originate from a trusted domestic source rather than a foreign one. This distribution method was key to its temporary credibility.
17. The website thispersondoesnotexist.com, launched in February 2019, used which architecture?
Correct. The site used NVIDIA's StyleGAN2 to generate photorealistic faces of non-existent people on demand. Its cultural impact was significant: it made GAN capability viscerally legible to a general audience who had no prior exposure to synthetic media research.
The site used NVIDIA's StyleGAN2. It was significant culturally because it made GAN-generated face synthesis immediately legible to a general audience — each page refresh producing a new photorealistic face of a person who does not exist.
18. What happened to deepfake detector accuracy on the FaceForensics++ benchmark when models were tested on cross-dataset content?
Correct. Models achieving 99%+ on FaceForensics++ dropped to 65–75% on cross-dataset evaluations — revealing that high benchmark accuracy doesn't predict real-world performance.
Cross-dataset evaluation showed dramatic accuracy drops (99%+ to 65–75%), exposing that detectors were overfitting to the specific artifacts of their training distribution.
19. What makes a description-only label (placed in a video's text description) ethically insufficient for synthetic media disclosure?
Correct. Content decontextualization is the core problem: a clip extracted from a labeled video and re-shared elsewhere carries no disclosure at all. Robust disclosure must be structurally inseparable from the content.
The core technical failure is decontextualization: descriptions stay on the original post while extracted clips circulate freely. Effective disclosure must travel with the content — not remain attached to a specific upload location.
20. What percentage of indexed deepfake videos were non-consensual intimate imagery according to the 2023 Home Security Heroes report?
Correct. 98% — a figure that surprises many who expect political or financial fraud deepfakes to dominate. NCII is by far the largest documented deepfake harm category by volume, with women comprising 99% of identified targets.
The figure is 98%. NCII is the dominant deepfake use case by volume — not political manipulation or financial fraud. Women comprise 99% of identified targets in the same report.