Punitive measures against creating, spreading deepfakes can deter them: CUTS

AI content checks: Sanctions on the creation and spread of deepfake content and misinformation can deter them, according to global think tank Cuts International.

CUTS International, Director for Research, Amol Kulkarni called for the deployment of technology interventions to suppress the abuse of AI-generated content, adding that internet users would need enough avenues to validate the legitimacy of content.

This is even more important as we approach elections, and underlines the role of dependable fact-checkers and legitimate flaggers. 

While the government’s March 15 advisory eliminates the need for permission for undertrial AI models, it goes on to stress for companies to disclose information to users for taking correct decisions on the Internet. 

"Though transparency is good, information overload and 'pop-ups' across user journeys may reduce their quality of experience. There is a need to balance the information requirements, with other implementable technological and accountability solutions which can address the problem of deepfakes and misinformation," Kulkarni said. 

The Government on March 1 released a directive to social media platforms to mark out under-trial AI models and avoid hosting illegal content following contention over Google’s AI platform with respect to a question about Prime Minister Narendra Modi. 

In the advisory issued to intermediaries, the Ministry of Electronics and Information Technology said non-compliance would result in criminal action.

A revised advisory on March 15 was issued by Meity, which eliminated the requirement for government approval for AI models which were not tested and under development.

The Ministry nevertheless stressed on the labelling of AI-generated content and information to users pertaining to the content generated being potentially unreliable.

Tackling the challenge of deepfakes and misinformation calls for clarifying the liability of all stakeholders in the internet ecosystem, from developers, uploaders, disseminators, platforms and consumers, Kulkarni noted.

"Penalty provisions for the development and dissemination of harmful deepfakes and misinformation could also create a deterrent effect. Technological solutions to tag potentially harmful content and shifting the burden on developers and disseminators to justify the use of such content could also be designed," he said

(With PTI Inputs)