Technology

Britain to ban non-consensual deepfake porn sharing

The British government has stated its intent to make it a crime to distribute pornographic deepfakes without the subject's consent.

The British government has stated its intent to make it a crime to distribute pornographic deepfakes without the subject’s consent.

The new crime is expected to be included in the massive and contentious Online Safety Bill, which will revamp the laws for policing harmful internet content in the United Kingdom. The bill’s approval this year was slowed by the current political instability, but the UK government hopes to bring it back to parliament in December for more discussion. The government declared that the new regulations would criminalise deepfakes and “downblousing” (capturing explicit photographs of a woman’s cleavage without her permission).

The government has announced its intention to crack down on “intimate image abuse,” including revenge porn, and non-consensual deepfakes are included in this broader effort. Problematic deepfakes are defined as “fabricated intimate photographs” that are then distributed without permission, but no further technical details are provided. One in fourteen adults in England and Wales report having been threatened with the disclosure of private photos of themselves.

Towards the end of 2017, pornographic deepfakes created with machine learning algorithms began to appear online. User-made pornographic videos featuring the likenesses of celebrities and women the users knew fast went viral on online communities like Reddit. Deepfakes were prohibited by certain popular online communities and pornographic websites in reaction, although the technology is still widely used in less-visited parts of the web. Despite widespread instances of these photos being used to degrade, assault, and frighten women, few jurisdictions have passed laws prohibiting their distribution. Currently, three US states (Virginia, Texas, and California) have legislation that references deepfakes.

Even though the initial round of pornographic deepfakes relied on AI techniques that superimposed targets’ faces onto pre-existing video clips, advancements in technology, such as text-to-image AI models, have made NSFW deepfake images much simpler and quicker. Furthermore, the line between cartoonish and photorealistic renderings is becoming increasingly porous due to technological advancements, which may further complicate legislative efforts to address the issue. When does a deepfake start to look like a particular person?

Professor Penney Lewis of the Law Commission (an independent organisation tasked with reviewing the laws of England and Wales) expressed her satisfaction with the government’s proposed revisions in light of the UK’s efforts to counteract this threat.

A new set of crimes has been created to ensure that more people who engage in abusive behaviour are brought to justice. The Guardian reported this. Lewis, who reportedly suggested the move, said, “Taking or disseminating intimate photos of a person without their agreement can inflict lifelong damage.”

The approval of the Online Safety Bill has been applauded by a wide variety of groups, particularly those concerned with child safety, who argue that more stringent regulations are necessary to safeguard Britons while they browse the web. Free speech advocates have voiced concern that the bill’s provisions targeting “legal but harmful content” could chill legitimate expression.