India’s Ministry of Electronics and Information Technology (MeitY) recently released an advisory, emphasizing stringent regulations for AI technology still in development. According to the advisory, any AI technology in its developmental phase must first secure explicit approval from the government before its release to the public. Additionally, developers are mandated to disclose any potential shortcomings or unreliability associated with the output generated by these technologies.

Furthermore, the advisory outlines plans for the implementation of a "consent popup" mechanism, aimed at informing users about potential defects or errors that may arise from AI usage. It also stresses the importance of labeling deepfakes with permanent unique metadata or identifiers to mitigate the risk of misuse.

In a move to ensure fairness and integrity, the advisory mandates that all intermediaries or platforms must guarantee that any AI model, including large language models (LLM), does not exhibit bias, discrimination, or pose a threat to the integrity of the electoral process.

However, these stringent measures have sparked criticism from some industry figures, who argue that they could stifle innovation and impose excessive burdens on companies. One critic expressed concern over the requirement for government approval even for deploying basic AI models, suggesting that it could hinder technological progress.

Developers are given a 15-day window to comply with the advisory, which may include undergoing stress testing or providing a demo for government officials after applying for permission to release a product.

While the advisory is currently not legally binding, it serves as a clear indication of the government's expectations and hints at the direction of future regulation in the AI sector. IT minister Rajeev Chandrasekhar emphasized that platforms offering AI must take full responsibility for their actions, suggesting that this stance will eventually be enshrined in legislation, reflecting a broader trend towards stricter accountability in the AI industry.