On This Day in AI: January 6, 2020
On January 6, 2020, Facebook announced from Menlo Park, California, that it would remove certain deceptive videos created or altered with artificial intelligence. Monika Bickert, the company’s vice president for global policy management, said the strengthened policy would cover media that appeared authentic but used AI or machine learning to merge, replace or superimpose content and make someone seem to say words they had not spoken. The decision arrived as election officials, researchers and technology companies warned that increasingly realistic “deepfakes” could be used to confuse voters during the 2020 U.S. presidential campaign.
The rule targeted a specific category rather than all misleading edits. A video had to contain changes that were not obvious to an average viewer, be likely to mislead about what a person said, and result from AI-based synthesis. Facebook explicitly exempted parody and satire, as well as video edited only to omit or rearrange words. Other manipulated posts could still be reviewed by independent fact-checkers. Material rated false or partly false could have its distribution reduced, carry warnings for viewers and be rejected as advertising, even when it did not meet the narrower test for removal.
Those boundaries drew immediate criticism. Reuters reported that the policy would not remove a widely circulated video slowed to make U.S. House Speaker Nancy Pelosi appear impaired because that clip used conventional editing rather than AI synthesis. Facebook said it had reduced the video’s reach after fact-checkers rated it false. The example exposed a central limitation: a crude alteration could deceive millions without qualifying as a deepfake, while sophisticated AI was not the only route to manipulation. The company’s decision therefore addressed an emerging technique but did not create a universal rule against deceptive political media.
The announcement was nevertheless an early attempt by a global platform to define AI-generated deception in enforceable terms. Facebook had already helped launch a Deepfake Detection Challenge with researchers and industry partners, and it partnered with Reuters on training for journalists who verified visual media. As synthetic audio, images and video improved, platforms later expanded labels, provenance tools and policies beyond the 2020 framework. The January 6 rule remains important because it captured both the urgency and difficulty of governing generative media: detection systems and written standards had to evolve together, but narrow technical definitions could leave harmful content untouched. It showed that AI safety was becoming a matter of public information systems and democratic trust, not only model performance.
The rule covered misleading, apparently authentic video altered with AI or machine learning to make a person seem to say words they did not actually say.
Facebook said it would also enforce other Community Standards when manipulated photos, audio or video violated rules on areas such as voter suppression or hate speech.
Parody and satire were exempt, as were videos changed only by omitting or rearranging words rather than using AI synthesis.
That distinction left some deceptive “cheapfakes” outside the removal rule even when fact-checkers found that viewers could be misled.
The policy forced a major platform to translate a fast-changing AI technique into practical standards for detection, review and appeal.
Later labeling and provenance efforts broadened the response, but the 2020 debate established that synthetic media could threaten trust well beyond the research lab.
Explore more of "On This Day ..."
Discover more events from the same date across news, politics, technology, sports, and other fields. Each link highlights significant moments that shaped history on different fronts.