New Delhi: Google is crafting a policy to educate creators on the proper use of ‘synthetic’ or deepfake content, as well as how to prevent its misuse. The policy will be regulating the use of artificial intelligence (AI)-generated content on platforms such as YouTube.
According to the company, the policy is focusing on creator disclosures when it alters reality, and mandates labelling of electoral advertisements crafted by GenAI, among others, besides enabling creators to watermark their content using Google’s tools.
“We are working on the right way to tackle this issue. On YouTube, we plan to have disclaimers about deepfakes or synthetic content for video descriptions, and in certain more sensitive cases, within videos themselves,” said Saikat Mitra, vice-president and head of trust and safety for Asia-Pacific at Google, in an interview with Mint.
“YouTube’s also working on policy (which will) expect content creators to disclose that this content has synthetic media and altering reality, and the consequences of not following that,” he added.
While Mitra did not elaborate on the punitive measures he said that as per existing policies accounts have been suspended and content taken down if it defies YouTube’s compliance guidelines.
Concern over synthetic content gained prominence after the targeting of public figures, including Prime Minister Narendra Modi and actor Katrina Kaif, through deepfake videos across social media platforms. On 23 November, union information technology (IT) minister Ashwini Vaishnaw met industry representatives, including Google, to discuss regulations to specifically target AI-generated fake content.
The Centre has also started work on draft deepfake regulations, and expects industry stakeholders to submit actionable measures to prevent its misuse by the weekend. Vaishnaw said the new regulations could either be implemented through amendments to existing rules, or the government could pass a new law specifically for such deepfake content.
A nuanced understanding of synthetic content will be key in the process of regulating it, Mitra said. “The term ‘deepfake’ typically has a negative connotation, but AI-generated synthetic content can also be positive, empowering creators to propagate businesses on social media. This is why watermarking of content, at Google’s end, will be key to segregating the good synthetic content from the harmful, and regulate each subset accordingly.”
Existing policies will also be applied to regulate the space, he added. “Deepfakes have been there for years, probably even decades. But, creating them was difficult. Today, it has been democratized, and it is thus important to distinguish between creation and dissemination of deepfakes, since they’re two separate problems. For dissemination, we already have a few policies around misinformation, and nothing needs to change there—what changes is the technology to detect dissemination, because that has become much tougher with the advent of large language models (LLMs).”
Google, and Facebook parent Meta are likely to be at the forefront of India’s upcoming deepfake regulation, industry stakeholders said. The onus on the regulation will be to try and find the creators of harmful deepfake content, which is the real challenge behind this entire process, Ashish Aggarwal, vice-president, public policy at industry body Nasscom, had said last month.
Google has enforced a global policy for advertisers to clearly disclose deepfakes for electoral content. Mitra said. Spreading of misinformation and AI-driven content have been hotly debated subjects for technology firms, particularly Google and Meta, over the past 50 years or so.
Milestone Alert!Livemint tops charts as the fastest growing news website in the world 🌏 Click here to know more.
Download The Mint News App to get Daily Market Updates & Live Business News.
Updated: 29 Nov 2023, 10:53 PM IST