INFORMATION TECHNOLOGY
European Union proposal: a watermark to identify and distinguish AI-generated content.
With the evolution of generative models, it is now increasingly difficult to distinguish between content created by humans and content created by Artificial Intelligence, whether text or images. For this reason, the European Union would like to introduce a watermark, a kind of "stamp" of recognition, to more easily distiguish the latter from the former.
This initiative sees an unprecedented collaboration between the authorities of the Old Continent and those of the United States, with the intention of requiring the main online platforms that use AI (Google, Meta, Microsoft, Amazon with Twitch..) to "mark" at the origin any element that has been generated by a chatbot such as ChatGPT or a text-to-image service such as Midjourney.
In this way, users should have the ability to immediately understand the nature of the content they are reading or viewing. This should therefore be a useful tool against disinformation campaigns that could receive a particularly relevant boost through the contribution of solutions such as GPT or similar.
Thierry Breton, the EU's internal market commissioner, and Vera Jourova, the European Commission's vice president for values and transparency, are reported to have described watermarking as a "track" to be included in the code of conduct against disinformation that should be observed to protect users from potential AI risks.
The commissioners' idea could be implemented in a variety of ways, not least because so far no specifics have been provided regarding what the watermark might look like. If it were to be visible, it could be more of a limitation for creativity and productivity through AI than for the generation of fake news.
(Source: MRW website - Author: Claudio Garau - Content Ownership: MRW)
This initiative sees an unprecedented collaboration between the authorities of the Old Continent and those of the United States, with the intention of requiring the main online platforms that use AI (Google, Meta, Microsoft, Amazon with Twitch..) to "mark" at the origin any element that has been generated by a chatbot such as ChatGPT or a text-to-image service such as Midjourney.
In this way, users should have the ability to immediately understand the nature of the content they are reading or viewing. This should therefore be a useful tool against disinformation campaigns that could receive a particularly relevant boost through the contribution of solutions such as GPT or similar.
Thierry Breton, the EU's internal market commissioner, and Vera Jourova, the European Commission's vice president for values and transparency, are reported to have described watermarking as a "track" to be included in the code of conduct against disinformation that should be observed to protect users from potential AI risks.
The commissioners' idea could be implemented in a variety of ways, not least because so far no specifics have been provided regarding what the watermark might look like. If it were to be visible, it could be more of a limitation for creativity and productivity through AI than for the generation of fake news.
(Source: MRW website - Author: Claudio Garau - Content Ownership: MRW)
