India Set to Impose Stricter Rules on AI Companies to Prevent Bias: Government Sources

India is set to implement stricter rules on companies working with artificial intelligence (AI) in order to prevent bias, according to government sources. The Ministry of Electronics and Information Technology is expected to amend the Information Technology (IT) Rules of 2021 to regulate AI companies and generative AI models. The amendments will require platforms using AI algorithms or language models to train their machines to ensure that they are free from any form of bias. This means that AI algorithms or language models trained on biased datasets or with inherent biases will be prohibited from being launched for open use. In addition to addressing bias in AI and language models, the amendments may also introduce regulations for deepfake and synthetic content, as well as explicit instructions for platforms dealing with loan apps. The government aims to expand the parameters for preventing bias beyond caste, religion, community, and national security. It plans to consult with experts to identify and address additional parameters that contribute to bias in order to enhance safety measures. The government will require AI models to undergo sandbox and stress testing to ensure the absence of biases before they are introduced to the market. These developments are part of the government’s larger efforts to introduce the Digital India Act, which aims to provide a secure and accessible internet for Indian users while reducing the dominance of major tech corporations. The Digital India Bill, which will replace the 23-year-old IT Act of 2000, places responsibility on intermediaries for the content published on their platforms instead of providing safe harbour provisions. However, the Digital India Act is not expected to be finalized before the general elections in 2024. Overall, these new regulations demonstrate India’s commitment to addressing bias in AI technologies and ensuring the trust and safety of the internet.