Last-48-hour hostile deepfake
A fabricated scandal drops too late for a correction to reach the same audience.
AI Risk Atlas Prototype/Demo — Unofficial independent experiment. Not an official xAI product. Scores can be wrong.
Owner · Election authorities, platforms, campaigns
Statement (NASA form)
Given that candidates, parties, and platforms are already deploying or failing to contain deepfakes at campaign scale, and state law is lagging live ads, there is a possibility of voters treating synthetic audio or video as authentic political reality resulting in illegitimate election outcomes, or legitimate ones that a large public no longer accepts.
- Condition
- candidates, parties, and platforms are already deploying or failing to contain deepfakes at campaign scale, and state law is lagging live ads
- Departure
- voters treating synthetic audio or video as authentic political reality
- Impact
- illegitimate election outcomes, or legitimate ones that a large public no longer accepts
Experimental share of compiled public capital that names this risk. Not a certified residual.
4 public sources · OECD AIM · AIID / The Verge · AIID / NTV
The civic risk is not only a hostile deepfake in the last 48 hours. It is also the authorised clone: a candidate who campaigns as a video that is not them, at a scale no human can audit. Platforms have already had to apologise to governments.
Simple upstream → via → downstream notes. Not a causal graph. Experimental.
Assumptions · EU labels are treated as landed; US and most other jurisdictions are not.
Override is stored on this desk only. It does not make the score official.
Each scenario has its own likelihood and consequence. The risk takes the most severe cell. Residual applies implemented mitigations to every scenario, then re-ranks.
A fabricated scandal drops too late for a correction to reach the same audience.
Voters cannot tell which appearances were the candidate. Accountability dissolves.
After enough synthesis, authentic footage of a concession or a count is treated as another deepfake.
Meta apologised to the Indian government over deepfake content and platform-operation failures during a live democratic cycle.
Candidates have used audio and video clones of themselves to reach voters who did not know they were talking to a synthesis.
US state deepfake laws are already lagging AI political advertising that is running in the same cycle.
X posts on the desk that evidence this risk. A signal can contribute to more than one risk.
Meta apology to India over deepfake and operations failures.
State deepfake laws lagging live AI political ads.
Candidates deploying self-deepfakes at electoral scale.
Deepfake of India’s home minister ran as a paid Facebook ad.
Anthropic watermarks and C2PA under Article 50.
Residual assumes only items marked in place. Highlighted rows are the remaining work needed to reach a composite of 12.
C2PA-style signing from camera to broadcast, with a visible break when the chain is missing.
Platforms and broadcasters · expedited 6 weeks · normal 5 months · −1 L · −0 C · −0 U
Ban undisclosed synthetic political content in a defined pre-election window, with real penalties.
Election lawmakers · expedited 2 months · normal 8 months · −1 L · −1 C · −1 U
If a campaign uses a clone, every impression carries a disclosure the platform cannot crop away.
Campaigns and platforms · expedited 2 weeks · normal 2 months · −0 L · −1 C · −0 U