Ali Video Watermark Passes Hollywood Authoritative Certification can be used for in -depth falsification of video traceability

Have you ever experienced a "friend" fraud experience? Mr. Chen reported to the police station and said he was cheated by "friends" for 50,000 yuan. After verification by the police, it turned out that the scammer collected the dynamic video information posted on social media on social media, and then synthesized a borrowing video to scam Mr. Chen. Such cases are not a small number of cases. Last year, the Anhui police seized a case of illegal use of deep synthesis technology to pseudo mobile phone users' dynamic videos to solve the identity verification and provide cases such as registered virtual mobile phone cards such as the black and gray industry.

At present, a stable and reliable new technology can be used to trace the source of fake videos and can identify the tampered areas. Recently, Ali Security announced that Ali Hidden Video Watermark 2.0 passed more than 160 attack tests of Cartesian, the preferred video watermark assessment and test company in Hollywood. It is the only video watermark technology that has passed Cartesian watermark twice in the past three years. Video traceability, data security, and protection of industry.

Ali Hidden Video Water Printing twice passed Cartesian certification of the preferred video watermark assessment and test company on Hollywood.

According to a report released by Tsinghua University in 2022, the number of newly released in -depth synthetic videos in 2021 has increased by more than 10 times on the mainstream and video websites and social media platforms at home and abroad. The application scenarios of deep synthesis are constantly enriched, such as repairing historical photos, AI sign language anchors, virtual idols and other applications. However, as the threshold of deep synthesis technology decreases, illegal acts and examples of fragrance, defamation, fraud, and extortion of false videos through deep synthesis technology have appeared in the society. At the same time, the release of the "In -depth Synthetic Management of Internet Information Services (Draft for Soliciting Opinions)" in 2022, which can be required to be identified and traced on deep synthetic videos.

Senior Security Expert of the Gemini Laboratory of Alibaba Safety Department introduced that the scene of traceability of the synthetic video can integrate the invisible video watermark into the synthetic video software. Whenever the user uses software to generate synthetic video, the video will be added at the same time to add simultaneously add One is used to identify the user's invisible watermark, and the invisible watermark does not affect the visual effect of the video. With the dissemination and circulation of the video, the watermark has been retained. Effect.

"We can also add a layer of invisible vulnerable watermark on the video that needs to be protected. When the video containing watermarks is tampered with, the fragile watermark on the video will also be destroyed. Detectors can be destroyed according to the watermark. Location recognition is tampered with areas, and effectively curb the abuse of in -depth falsification technologies such as Deepfake. "Du Ming added.

The road is one foot high and one foot high, and there will always be black and gray products trying to destroy the extraction of watermark technology to achieve the goal. For example, video editors introduce videos to rotate, zoom, translate, or introduce signal processing attacks to change brightness and saturation and saturation , Comparison, etc., attempts to make people unable to correct watermarks from the video.

In order to resist stronger malicious attacks and wider attack methods, the Alibaba Water Printing Algorithm personnel have covered the attack scene under more than 160 extreme conditions proposed by Cartesian after nearly a year of algorithm development and continuous testing. Facing the robustness of multiple attacks.

"With the continuous decrease in the technical threshold and joint attack threshold of video professional editing, video watermark technology will also face many new risks. Hollywood certification companies also design new tests for new attacks each year. These test projects are confidential. Invisible video watermarks are also performing well when blind testing these projects. "Du Ming introduced.

Since 2019, Ali Safety is committed to building independent and controllable domestic top and safe watermark technology. Invisible video watermark technology has passed the robust test of many authoritative institutions at home and abroad. Du Ming emphasized that the first -top technology of self -developed technology is to reduce the operating costs of long videos, and the other is to meet the requirements of copyright protection and data security in the digital scenarios.

Du Ming emphasized: "The security issues faced by new technologies also urgently need digital watermark technology to give corresponding solutions, such as security needs in the fields of driverless and artificial intelligence. Solving social problems will bring people more sense of security. "

- END -