Google, OpenAI Struggle To Perfect Watermarking System for AI-Generated Content Amid Deepfake and Misinformation Concerns: Report
Google and OpenAI are facing challenges in developing a reliable watermarking system for AI-generated content to combat deepfakes and misinformation, according to a report.
New York, November 27: Google and OpenAI, two of the leading AI companies in the world, are struggling to find perfect ways to "watermark" their content to identify the source of origin. Several artificial intelligence companies have been trying to attempt to flag their content to identify non-human-generated text. AI has become a powerful tool and feature for many to get things done faster, such as text or image generation, finding answers, or solving complex math problems.
According to a report by Bloomberg, OpenAI and Google have been struggling to flag or watermark their content for two years to know whether humans or AI-generated the text. Sam Altman-run OpenAI announced it was working on a "Watermarking" system to identify the AI-generated text. However, there is no further update from the company about this feature. iPhone 17 Pro Likely To Include Rectangular Camera Bump, Revamped Design Compared to Current Models: Report.
The report mentioned that OpenAI called this method "highly accurate," but it has yet to announce or implement it. The company said that flagging AI text could affect non-native English speakers who rely on AI tools for writing, while many users. Bloomberg said that an OpenAI spokesperson highlighted that the company had taken necessary steps considering the complexities involved and impacting the broader.
Adding a watermark to the AI-generated video or text would be necessary for identifying the tools and spread of deepfakes and misinformation. Some countries have already started introducing laws to curb the control of misinformation and harm to the affected parties due to AI-generated content. Also, marking the AI content would help the company identify the human-made work. OpenAI Signs 1st MoU With Korea Development Bank To Foster Growth of AI Ecosystem in South Korea.
The report mentioned that the companies have been pushing cryptographic signatures to identify if someone tempered with the original content and help trace its editing history. However, people can easily remove such watermarks by taking screenshots of the content.
(The above story first appeared on LatestLY on Nov 27, 2024 01:33 PM IST. For more news and updates on politics, world, sports, entertainment and lifestyle, log on to our website latestly.com).