Deepfakes regulation to target both creators and social media platforms

New Delhi: India is gearing up to crack down on AI-generated deepfakes and misinformation with a new regulation that could impose financial penalties on social media platforms enabling the spread of such harmful content, along with the creators. The government, along with other stakeholders, aims to strengthen the reporting mechanism within 10 days to empower citizens to combat AI-generated harmful content on the internet, as stated by the Minister of Information Technology and Telecommunications, Ashwini Vaishnav.

Deepfakes pose a new threat to democracy, weakening trust in society and its institutions, according to the minister. He mentioned that financial penalties could be included in the regulation, stating, “When we regulate, we need to impose penalties not only on the person uploading or creating but also on the platforms.”

Minister Vaishnav discussed tackling deepfake content in a meeting with representatives from technology giants such as Meta, Google, and Amazon on Thursday. He emphasized the need for social media platforms to be more proactive, considering the rapid spread of damage from deepfake content, which can go viral within minutes. Strengthening public trust in society is crucial for the protection of our democracy, he added.

Earlier, Mint reported that the government aims to regulate deepfake content, urging social media platforms to scan and block deepfakes. Vaishnav emphasized that social media platforms need to be more responsive, recognizing that damage from deepfake content can occur rapidly, and delayed responses may not be effective.

Consensus has been reached on clear, actionable items in the next 10 days based on four key pillars: identifying deepfakes, preventing the publication and viral sharing of deepfakes and profound misinformation, strengthening the reporting mechanism for such content, and raising awareness through collaborative efforts between the government and industry, Vaishnav stated.

Deepfakes refer to synthetic or algorithmically altered media presented digitally to mislead or replicate someone in a misleading manner, often utilizing artificial intelligence. The new regulation could be presented either as an amendment to India’s IT rules or as an entirely new law.

“We can regulate this space either through a new standalone law, or as amendments to existing rules, or through a new set of rules under existing laws. The next meeting is scheduled for the first week of December,” Vaishnav added.

In this discussion, we’ll be talking about the regulation of deepfake, which will later be opened for public consultation,” said Vaishnav.

The minister mentioned that under the Information Technology (IT) Act, platforms won’t be subject to the ‘safe harbor protection’ until they swiftly take substantial actions.

During Thursday’s meeting, other aspects discussed included the issue of AI bias and discrimination, and how the existing reporting mechanism could be modified.

Last week, the government issued notices to social media platforms after reports of deepfake content. Concerns have escalated over deepfake videos, especially after high-profile public figures like Prime Minister Narendra Modi and actress Katrina Kaif became targets.

On Wednesday, during the virtual G20 Summit, Prime Minister Modi also addressed the issue of deepfakes, emphasizing the need for international cooperation to tackle the challenges posed by this technology.

Stakeholders in the industry were generally positive about the discussion on regulating deepfakes during Thursday’s meeting.

As part of the consultation, a spokesperson from Google, who was part of the meeting, mentioned the company’s efforts in developing tools and systems to prevent the misuse of technology and to enable users to better assess online information.

“We have long-standing, robust policies, technology, and systems in place to identify and remove harmful content on our products and platforms. The company is launching new products operated by generative AI while applying this norm and perspective,” stated the company in a statement.

Meta did not provide immediate answers to the questions raised.

Ashish Agrawal, the head of public policy at the software industry body NASSCOM, highlighted that while India already has laws to penalize offenders for misinformation, strengthening the rules for identifying creators of deepfakes is crucial.

He stated, “A more significant discussion is how to catch the 1% of malicious users creating deepfakes – this is an identification and enforcement problem that is before us.”

Technology today can help identify synthetic content. However, the challenge lies in distinguishing harmless synthetic content from harmful ones and promptly removing them. An upcoming tool, currently under extensive consideration, embeds watermarks or labels in all digitally altered or generated content. This alerts users about synthetic material and associated risks, while also empowering them to report it promptly.

A senior industry official familiar with the matter mentioned that most companies have adopted a “compliance-friendly approach.” However, addressing the challenges of filtering deepfakes and false information on platforms with a large amount of non-English content can be daunting. It is also crucial for such platforms to handle sensitive information responsibly.

Rohit Kumar, co-founder of the policy think tank The Quantum Hub, emphasized the need for regulation to be mindful of the cost of enforcement. He suggested that while defining responsibilities, platforms should be encouraged to adopt a nuanced approach to reduce the compliance burden. For example, setting a “virality” threshold could be established, and platforms could be instructed to prioritize reviewing and removing content that starts to go viral.

He emphasized that cybersecurity should not be weakened, as the responsibility for the damage caused by deepfakes should rest with the individual creating and posting the video, not the platform.

For more latest news click

Leave a Reply

Your email address will not be published. Required fields are marked *