TECH
SynthID was announced in August last year and has since served as a watermark to identify artificially generated images. During the I/O event, this Tuesday (14), Google announced that the feature will be expanded to two more formats: videos and texts generated with AI.
The announcement was made by the CEO of Google's DeepMind, Demis Hassabis, who also spoke about the big tech's new video generator, Veo.
SynthID emerged last year as a way to create images using AI without them being identified by humans, but rather by systems. Now, the tool has received updates to expand the identification of artificially generated content for videos and texts.
SynthID also has a hand in music: it allows the injection of watermarks into songs artificially generated by the Lyria program, without them being identified by the human ear.
Protections against technology...Watermarks have become a trend among technology companies involved in the development of AI;
The intention is to allow the identification of artificially generated content and limit disinformation, especially when this content is being used for malicious purposes in electoral campaigns, in political and humanitarian contexts, and in health;
For example, some time ago, AI deepfakes showed US President Joe Biden, hospitalized, at the Mexican border welcoming Mexican immigrants and alongside a body double.
The inclusion of watermarks to identify content generated by AI is one of Biden's proposals for regulating technology in the country.
mundophone
No comments:
Post a Comment