The Maharashtra government, along with IIT Bombay, built an AI-based tool that helps to identify suspected illegal Bangladeshi and Rohingya migrants residing in the state. The AI tool will help study how a person speaks, including accents, tone, and specific linguistic patterns, to identify individuals who do not have valid Indian documents.
The AI tool will help the authorities serve as a support system rather than just a final proof of nationality. The plan is still in process, which has triggered a mix of political messaging, public debate, and questions about rights and accuracy.
The project was released after Maharashtra Chief Minister Devendra Fadnavis mentioned the tool during the election manifesto. Fadnavis assured that the government’s promise of “free Mumbai,” saying that technology can play a key role in identifying illegal migrants.
According to the sources, the AI model has been under testing for some time, but is still far from perfect. The results obtained early on suggest about 60 per cent accuracy.
The government believes that the project will flourish, improving data quality and enabling more stringent requirements to be met. Civil rights activists and tech experts have raised red flags, warning that language-based identification can easily go wrong.
The activists also pointed out that many Indian citizens, from West Bengal, Assam, and border regions, are identical to Bangladeshis. They suggest that this could lead to innocent people being wrongly flagged.
There have been privacy concerns. Questions are being asked about how the voice samples will be collected, and whether people will be aware that they are being paralysed. The government has not provided detailed answers to these questions.
Apart from the AI initiative, the state indicated plans to set up detention centres for the undocumented migrants before deportation.
