It’s no secret that AI use within the gaming industry is rapidly evolving, and studios around the globe are using the technology to streamline game development. While some are still opposed to the technology, the industry is exploring interesting ways of integrating AI tools within the game development process.
Sony is no different, and its latest patent hints at revolutionary AI use for censorship. The Clawsomegamer team’s latest finding shows that Sony has filed a patent for automated, bespoke content editing using AI, primarily aimed at censorship and tailored experiences for users sensitive to certain elements in games.
The patent, filed by Sony Interactive Entertainment on 9th December, outlines the use of a dedicated processor to access data from audio and video. According to the patent, a processor is programmed to execute the AI model to identify content that falls within user-specified parameters, and a second processor then obfuscates the identified content.
Sony notes that this patent primarily relates to computer ecosystems, citing PlayStation and Microsoft or Nintendo consoles as targets of this technology. The description highlights how this AI model could be used to mute specific scenes with slurs. In certain cases, the censored audio could be replaced with an entirely different piece of content through deepfake or other technology.
The AI will also be able to identify specific words that users deem offensive. Moreover, Sony Interactive Entertainment’s patent cites examples like nudity, alcohol use, and clowns with red balloons as targets of censorship that would be identified and obfuscated by the AI.
As for how the user will interact with such a system, the patent outlines that users could be informed of incoming content both ahead of the playback and during playback itself. How this selection would be carried out is highlighted in an example in Figure 4 below. Then, Figure 5 shows the series of steps taken by the AI to carry out the censorship process.
The user can then choose from options like “remove the content” or “replace the content.” Also, because the AI system aims for a highly flexible solution, each user can have custom parameters for what they consider acceptable or unacceptable, even when looking at content or language that is generally deemed offensive.
An example of the UI used for such a system is shown in Figure 7 below, showcasing how users can customize the obfuscated content to their liking.
Sony ultimately aims to offer a system that dynamically adjusts content on the fly and adapts to modern standards. For instance, in cases where standards for PG-13 ratings change over the years, the AI could modify content on the fly and ensure it remains safe for users of all ages.
Because age ratings and content appropriateness have been major topics of discussion within gaming in the past, this patent could essentially solve these problems altogether and make any game playable for gamers of all ages.
