LI Network
Published on: January 8, 2024 at 00:10 IST
The government is reportedly considering changes to the Information Technology (IT) Rules of 2021 to regulate artificial intelligence (AI) companies and generative AI models.
Sources familiar with the matter suggest that these amendments will likely enforce stringent regulations on platforms utilizing AI algorithms or language models for machine training to ensure they are devoid of any form of bias.
Expectedly, the Ministry of Electronics and Information Technology is anticipated to announce these amendments soon. The proposed rules aim to prohibit AI algorithms or language models from operating if they exhibit inherent biases or are trained on biased datasets, ultimately restricting their open use.
Additionally, alongside measures targeting bias prevention in AI and language models, the anticipated amendments might introduce regulations concerning deepfake, synthetic content, and specific guidelines for loan apps.
An anonymous senior government official explained, “The existing bias parameters revolve around factors like caste, religion, community, and national security. Platforms will have a legal obligation to adhere to these parameters, expanding internet trust and safety.”
The government plans to conduct consultations to identify further parameters eliminating biases, ensuring a safer online environment. Stressing the importance of testing, another official highlighted the necessity for sandbox and stress testing of AI models before their market release to eradicate biases.
Earlier reports suggested the potential inclusion of provisions in the forthcoming Digital India Bill empowering users to demand algorithmic accountability from internet intermediaries and social media companies. However, the bill’s release might be postponed until after the general election.
The decision to amend the IT Act reportedly stems from the rapid advancements in generative AI while the Digital India Bill is still in progress. The proposed amendments are expected to offer precise guidelines addressing deepfake, synthetic content, and fraudulent practices in loan apps.
The government’s approach towards preventing fraudulent loan apps from functioning in India could involve strategies like white-listing legitimate applications or implementing a banking gateway to restrict transactions through unauthorized apps or individuals.
The official emphasized, “Cutting off access to legitimate banking solutions for operators of fraudulent apps through predictive algorithms could significantly hinder their operations.”
By fortifying regulations and utilizing predictive algorithms, the government aims to curb the operations of fraudulent platforms, safeguarding users from potential risks associated with biased AI models and unapproved loan apps.