India’s new Data Protection Bill to shield personal data from AI

According to the most recent draft of the Digital Personal Data Protection (DPDP) Bill, 2023, which was approved by the cabinet earlier this month, generative artificial intelligence (AI) platforms like ChatGPT or Google’s Bard may not be able to process the personal data of Indians available in the public domain.

Of course, the final bill’s draft is not yet available to the public. To conduct this research, experts used a leaked copy of a manuscript that has been discussed in legal and policy circles. Due to this, the majority chose not to give their names.

In the most recent version, a clause that had previously allowed search engines to process personally identifiable information that was made publicly available has been removed.

If generative AI systems scrape the Internet to use such data, they may expose themselves to litigation in the country, similar to the ones they are presently facing in the US, according to experts.

“This removal indicates that AI chatbots shall collect and process publicly available personal information only after obtaining consent from data principals at the commencement of its processing,” the expert said.

Removing Clause 8(8), which classified any ‘processing of publicly available personal data’ under public interest as a requirement for deemed permission, could have an influence on new AI developments like ChatGPT, according to a technology expert at a public policy think tank.

This provision was included in the DPDP Bill as it stood in 2022. The DPDP Bill, 2023 has been placed on the government’s agenda for the forthcoming monsoon session of the Lok Sabha for consideration and passage.

Experts noted that the US Federal Trade Commission (FTC) opened an inquiry into OpenAI, the firm that developed ChatGPT, last week to see whether it had violated consumer protection rules by scraping public data.

“This removal indicates that AI chatbots shall collect and process publicly available personal information only after obtaining consent from data principals at the commencement of its processing,” the expert said.

The FTC has questioned OpenAI extensively in a 20-page letter regarding, among other security issues, the startup’s handling of personal data and the training of AI models.

Other countries, like Italy, have put limitations on OpenAI. According to experts, the FTC action may establish a wider standard for how generative models like ChatGPT train their language models.

OpenAI CEO Sam Altman said in a tweet that the company will work with the agency. Its technology is safe and pro-consumer “and we are confident we follow the law”, he said. Altman also said the company protects user privacy and designs its systems “to learn about the world, not private individuals”.

India has openly stated that it is not currently considering regulating generative AI models. CIO Insider