L’Artificial intelligence it has already revolutionized the approach of millions and millions of users to the internet, also demonstrating dark sides that should not be underestimated.
In this context, phenomena such as the infamous deepfake, they are not dangerous as far as common online scams are concerned. Also the political worldIn fact, it could be overwhelmed by AI abuse, especially in conjunction with elections. In this regard, some giants in the sector such as Microsoft e Meta they seem to have broken the deadlock, adopting some solutions to protect politicians and voters.
The Mountain View giant, to help limit online misinformation, has decided to launch a new tool. It is a system of watermarkingdeveloped by Coalition for Content Provenance Authenticity (C2PA), capable of protecting political candidates. Through this solution, Microsoft guarantees them to protect their image and prevent deepfakes that portray their image from being created.
This service, which has been given the name of Content Credentials as a Service, goes to include some metadata in an image or video. These concern the origin and methods of creation of the content, also determining whether it is the result of the development of an AI tool.
Microsoft and Meta, different initiatives but aimed at curbing political deepfakes
As already announced, however, Microsoft is not the only company to have moved in this field.
Meta, during 2024, intends to apply a advanced control method to which advertisers who publish political ads should be subjected. Those who deal with advertising, therefore, will have to provide clear indications when content is manipulated in some way, even slightly.
Only some digital alterations such as changes to the image sharpnesspossible clippings o color adjustments are not covered by these new rules. Once any substantial changes to videos and images have been identified, they will be cataloged in a database of Meta and analyzed.
Speaking of which, Meta’s president of global affairs Nick Cleggstated how “Advertisers running these ads do not need to disclose when content is digitally created or altered in ways that are irrelevant or immaterial to the complaint, assertion or issue raised in the ad“.