Dana leads trust and safety for a creator-video platform. A filmmaker reports that someone has used their likeness in a fabricated intimate video. Dana does not need to display the video to recognize the stakes: a person needs protection, and the platform needs a reliable response.
YOUR MISSION
Follow Dana from report to a stronger workflow. You will explain two kinds of trust damage, recognize NCII and child-safety risks, identify the US platform response rules, apply the 48-hour sequence, and assemble six creator safeguards.
Educational, not legal advice. This US-focused lesson explains selected rules and responsible practices; it is not a complete compliance guide. Serious-harm topics are discussed without graphic examples.
DANA Trust-and-safety lead · fictional case
A report needs a response
01 / RECOGNIZE THE HARM
Trust can break in two directions
Source material → edited export. Verify origin and context.
01 · False belief02 · Misplaced doubt
BELIEVING AN INVENTED EVENT
A misleading deepfake can invent an event, distort public discussion, or make fabricated footage seem like evidence. In Dana's case, viewers may mistake the fabricated video for a recording of something the filmmaker actually did.
DOUBTING A REAL EVENT
There is a second effect: once convincing fakes seem commonplace, people may dismiss genuine recordings as fake too. This weakening of shared grounds for knowing what happened is epistemic erosion. Repeated deception can breed cynicism: “Why believe any footage?” Dana's goal is careful verification and context, rather than universal suspicion. This is an explanatory application of the information-integrity risks described by NIST, not a claim that all synthetic media produces these outcomes. [S3]
02 / CENTER THE PERSON
Invented scene. Real person.
A private report. A person who deserves a response.
NCII Nonconsensual intimate imagery
NCII means nonconsensual intimate imagery: intimate images or videos shared without the depicted person's consent. It can include synthetic depictions. A digital forgery can attach an identifiable person's likeness to an intimate scene that never happened.
The event may be invented; the person's distress, loss of control, reputational harm, and vulnerability to extortion can be real. Dana treats the report as a harm report, rather than dismissing it because the pixels are generated. Permission to film someone, or to use a normal portrait, is not blanket permission for intimate fabrication or distribution. [S1, S8]
Origin is a technical question. Consent is a human one.
Dana confirms that the reported intimate video is fabricated. Which conclusion should guide her response?
Fabrication does not remove the person's distress, loss of control, or exposure to extortion. Try again.
A label can clarify origin, but cannot supply the depicted person's consent. Try again.
Yes. The person can suffer real harm, while misleading imagery also distorts what audiences believe.
That blanket suspicion deepens epistemic erosion. Verify evidence and context instead. Try again.
03 / PROTECT BEFORE PUBLICATION
Protect minors before the next report
INDUSTRY POLICY
Dana tests whether the workflow would protect someone under 18. Easy creation tools make prevention urgent: sexualized synthetic imagery can target children without their participation. Industry safety policies prohibit child sexual exploitation, including simulated sexual activity or sexualized nudity; Google expressly prohibits AI-generated child sexual abuse material and nudification of minors. These are examples of industry rules, not a claim of one universal policy. [S6, S7]
Pause the release. Give a specialist the authority to act.
PUBLIC ENFORCEMENT2024 San Francisco
Public authorities have also acted against automated “nudification” services. San Francisco's city attorney brought a 2024 civil action alleging that such websites violated California law. A filed complaint states allegations, not a finding that every defendant broke the law. Dana strengthens prevention and routes suspected child exploitation to trained specialists; she does not ask ordinary creators to collect or circulate abusive material. [S8]
04 / KNOW THE RESPONSE RULE
A valid request starts the response
The TAKE IT DOWN Act addresses certain nonconsensual intimate depictions, including digital forgeries. The FTC enforces Section 3's platform notice-and-removal requirements; this is distinct from the Act's criminal provisions. [S1, S2]
WHO IS COVERED?
Covered platforms include services primarily providing a forum for user-generated content. Another statutory category covers services regularly publishing, curating, hosting, or furnishing nonconsensual intimate depictions. Scope has definitions and exclusions; not every website is automatically covered. Dana's fictional platform mainly hosts user videos. [S1]
A valid written request includes a physical or electronic signature from the depicted individual or an authorized representative; enough information to identify and locate the depiction; a brief good-faith nonconsent statement with relevant information; and contact information. A signature and a vague complaint alone are insufficient. Dana makes the process clear and accessible, including for people without platform accounts. [S1, S2]
05 / PRACTICE THE RESPONSE
One request. One 48-hour window.
For a valid request received Monday at 10 a.m., Dana's outside deadline is Wednesday at 10 a.m., not two business days. Remove the reported depiction as soon as possible, no later than 48 hours after receipt. Within that same window, make reasonable efforts to identify and remove known identical copies on the platform. [S1, S2]
VALID REQUEST RECEIVEDMon · 10:00 a.m.
ONE CONTINUOUS WINDOW
OUTSIDE DEADLINEWed · 10:00 a.m.
48hours maximum
Choose Dana’s next action0 / 4 connected
01Next action
02Next action
03Next action
04Next action
Start with the event that triggers the response clock.
TRY A DIFFERENT ARRIVAL
DEADLINE · LOCAL TIMEWeekends count. The clock never resets between steps.
The operational sequence is: receive the valid signed request → identify and locate the material → remove the depiction → identify and remove known identical copies using reasonable efforts. The clock starts on receipt of the valid request, not after Dana finishes internal review, and it does not restart for copies. Treat 48 hours as a maximum, not permission to delay. This is not a guarantee of deleting every altered version everywhere on the internet. [S1, S2]
A valid request reaches Dana on Monday at 10 a.m. The main depiction is removed Tuesday. What comes next?
Yes. Copy identification and removal belong within the same 48-hour window.
The clock does not restart after the first removal. Try again.
The platform must make reasonable efforts; the requester need not identify every duplicate. Try again.
Removing the reported depiction alone leaves the copy obligation unfinished. Try again.
06 / HARDEN THE WORKFLOW
Build the workflow before the next upload
Dana closes the immediate response and asks what creators can change before the next upload. The six safeguards work together; none substitutes for the others or establishes legal compliance by itself.
SAFEGUARD 1 OF 6
Provenance
1. Provenance via cryptographic watermarking — retain a verifiable origin trail. More precisely, use cryptographically signed provenance, such as Content Credentials, plus watermarking where supported. These are distinct techniques: a watermark is an embedded signal; a signature helps verify integrity and origin assertions. Neither proves that a scene is true or that its use is consensual. Dana asks for the source file and retained provenance record. [S4, S5]
DANA’S WORKFLOW GAP
Dana has an export but cannot trace its origin.
Which evidence would close this gap?
SAFEGUARD 2 OF 6
Written consent
2. Meaningful written consent — record whose likeness is used, the synthetic changes, intended context, distribution, and boundaries. If the proposed use changes, revisit permission. Dana asks for specific authorization rather than treating an ordinary appearance release as permission for any synthetic use. This is a recommended practice, not a universal statutory form requirement. [S1 supports the creation/publication distinction; workflow details are lesson recommendations.]
DANA’S WORKFLOW GAP
The release covers the shoot, but not the proposed synthetic change.
Which evidence would close this gap?
SAFEGUARD 3 OF 6
Disclosure
3. Appropriate disclosure of synthetic media — make a clear, contextual disclosure when audiences could otherwise mistake an invented scene for a recording. Dana checks how the disclosure survives export and sharing. A label does not cure nonconsent. [S4; workflow recommendation.]
DANA’S WORKFLOW GAP
The disclosure disappears when the clip is exported.
Which evidence would close this gap?
SAFEGUARD 4 OF 6
Protect minors
4. Protecting minors — block sexualized uses of minors and route uncertainty to trained specialists before publication. Dana checks tools, source material, and escalation ownership. Calling a depiction simulated does not make it acceptable under these child-safety policies. [S6, S7]
DANA’S WORKFLOW GAP
A flagged age-related use can still pass straight to publication.
Which evidence would close this gap?
SAFEGUARD 5 OF 6
Harm-reporting
5. Clear harm-reporting — make reports easy to find and route them to someone who can act. Dana assigns an owner and tracks receipt, deadlines, actions, and follow-up. An ignored inbox is not a functioning response process. [S2, S7]
DANA’S WORKFLOW GAP
The report inbox exists, but nobody owns the response deadline.
Which evidence would close this gap?
SAFEGUARD 6 OF 6
Human oversight
6. Continuous human oversight — assign reviewers authority to pause a release, review flagged outputs, respond to reports, and revise safeguards as tools change. Dana schedules follow-up after the incident rather than assuming one filter fixes the workflow. [S3, S7; workflow recommendation.]
DANA’S WORKFLOW GAP
The automated filter is the only release decision-maker.
Which evidence would close this gap?
Build all six safeguards to finish your workflow.
Your assembled workflow
Your completed safeguards will appear here.
Provenance
Written consent
Disclosure
Protect minors
Harm-reporting
Human oversight
07 / TAKE IT INTO YOUR NEXT PRODUCTION
The road ahead is shared
Dana's response links three jobs: recognize harm, act promptly, and prevent recurrence. Synthetic imagery can undermine trust and injure real people. A valid Section 3 request starts a single 48-hour window for the reported depiction and the required reasonable-efforts work on known identical copies. Six creator safeguards support prevention and accountability.
01
Platforms
Platforms operate reporting and removal systems.
02
Regulators
Regulators enforce applicable requirements.
03
Creators
Creators make choices about consent, context, disclosure, and release.
Shared responsibility means each party has work to do to protect human dignity; it does not transfer everyone's duties to the person reporting harm.
From one report to a stronger workflow.
Short glossary:
NCII
nonconsensual intimate imagery; intimate images or videos shared without the depicted person's consent, including synthetic examples.
Epistemic erosion
weakening of shared trust in evidence and the ability to know what happened.
TAKE IT DOWN Act
a US law addressing certain nonconsensual intimate depictions; Section 3 concerns platform notice and removal.
48-hour takedown
the outside response window after a valid request, with reasonable efforts to identify and remove known identical copies in the same period.
Human oversight
ongoing review by people who can intervene, resolve reports, and improve safeguards.
After Dana debunks the forgery, a viewer says authentic behind-the-scenes footage must also be fake. What does this illustrate?
Dana stress-tests the workflow with a report involving a sexualized synthetic depiction of a minor. Which response fits the lesson?
On Dana's user-video platform, which intake summary best supports a valid Section 3 request?
A valid request arrives Friday at 6 p.m. Which plan follows the response window?
Dana finds signed provenance and a synthetic label, but no written permission and no report owner. What should she recommend?
REFLECT / YOUR CASE REVIEW
Your response, reviewed
0%0 of 5
Your results
Review your decisions, then take these safeguards into your next production.
Pass threshold: 80% · Four of five decisions
01
After Dana debunks the forgery, a viewer says authentic behind-the-scenes footage must also be fake. What does this illustrate?
CORRECT RESPONSE
Epistemic erosion spreading doubt to real evidence.
Fake footage can also make genuine footage easier to dismiss.
02
Dana stress-tests the workflow with a report involving a sexualized synthetic depiction of a minor. Which response fits the lesson?
CORRECT RESPONSE
Escalate to trained specialists and enforce child-safety rules.
Minors require special protection; synthetic origin is not an exemption.
03
On Dana's user-video platform, which intake summary best supports a valid Section 3 request?
