OpenAI To Add Watermark to AI-Generated Using 'DALL-E 3 Model', Will Adopt 'C2PA Technical Standard' Amid Growing Concerns of Misusing Information Ahead of Election
OpenAI will start adding digital watermarks to its AI-generated images using DALL-E 3 model using the C2PA technical standard. Amid the growing misuse of the information, the company deems it fit to use this standard despite it is not a perfect solution.
Mumbai, February 7: OpenAI announced that it will add the watermark to the images generated using the DALL-E 3 model. The company published a post on its official website saying the watermarks would be added to the Content Provenance and Authenticity (C2PA) standard. By adding the watermark to the images, the ChatGPT developer can ensure that the photos were developed using AI, not humans, and add authenticity and trust to the digital information.
By adding a watermark to the images generated using DALL-E 3, the company could help the users identify images. The company posted that C2PA is a technical standard that "allows publishers, companies, and others to embed metadata in media for verifying its origin and related information." The company further informed that the C2PA is not only for AI-generated images but also the same standard that camera manufacturers, news organisations and others adopt. Meta Expands ‘Take It Down’ Programme to More Countries and Languages To Help Millions of Teens Combat Sextortion.
OpenAI Announces to Watermark AI Generated Images Using DALL-E 3:
The C2PA technical standard helps companies certify media content's source and history or provenance. In its official post, OpenAI said that the images generated with ChatGPT on the web and the company's API serving the DALL-E 3 model will include the C2PA metadata. The company further emphasised that adopting these methods will help in establishing "provenance" and encourage users to recognise the signals as key to increasing the trustworthiness of digital information. AI Co-Pilot Launched by Non-Profit Organisation Wadhwani Foundation To Boost Employability for 10 Million Indian Youth by 2027.
According to the report by Indian Express, one can still tamper with the watermarked images by cropping them or manipulating the metadata. The report suggested that the metadata will be removed when the users take screenshots of such AI-generated images or upload them to the social media platform. On the other hand, Business Insider reported that OpenAI knows these plans are not perfect; however, there have been growing concerns about spreading misinformation containing images, audio, videos, etc., before the upcoming elections.
(The above story first appeared on LatestLY on Feb 07, 2024 07:47 PM IST. For more news and updates on politics, world, sports, entertainment and lifestyle, log on to our website latestly.com).