government: Govt may amend IT Act to add new rules for AI, GenAI models

The amendments are likely to mandate that platforms which use artificially intelligent algorithms or language models to train their machines are free from “bias” of any kind, they said.

The amendments are likely to be notified soon by the Ministry of Electronics and Information Technology and mandate that such AI algorithms or language models that have inherent biases or are trained on such datasets which give out biased answers be barred from launching for open use.

Apart from the safeguards for preventing bias in AI and language models, the IT Rules amendments may also bring new rules for deepfake, synthetic content and explicit instructions for platforms on loan apps, said one of the persons.

“The parameters for bias that we have right now are centred around caste, religion, community and national security, among other parameters. There will be a legal obligation on platforms to ensure all these parameters are followed, thereby expanding the ambit of trust and safety of the internet,” a senior government official told ET on condition of anonymity.

Discover the stories of your interest

The official said the government will conduct consultations on additional parameters to eliminate bias apart from those already known and understood to ensure safety.

These AI models, another official said, must undergo sandbox and stress testing to ensure that such biases are not present before releasing the tool in the market.

ET had reported on September 29 that the government is likely to introduce in the upcoming Digital India Bill similar provisions that seek to empower users to seek algorithmic accountability from internet intermediaries and social media companies.

The Digital India Bill, however, may now be released only after the general election, according to people in the know. They said the government’s decision to amend the IT Act may be prompted by the fact that the Digital India Bill is still some time away and developments around generative AI are progressing rapidly.

Also read | Generative AI’s wild 2023: how the viral technology took over the world

“The advisory on deepfake and synthetic content will be very specific on the dos and don’ts for platforms and how to tackle the issue. For the loan apps, we want these intermediaries to take the responsibility of preventing any kind of fraudulent advertising or stop such apps which do not have the requisite approvals from being hosted altogether,” said the official cited earlier.

Since the IT ministry is the “custodian of safety and trust” on the internet, its approach to preventing such loan apps from operating in India is likely to be either through a white-listing of the legitimate applications or placing a banking gateway which prevents the transactions through such apps or people, the official said.

“An online betting or fraudulent loan app can only function if they are allowed to connect to a legitimate banking solution. If we ensure that such people who operate these loan apps are denied access altogether to the banking system by the use of predictive algorithms, a lot of these platforms will wither away,” the official said.

The amendments are likely to mandate that platforms which use artificially intelligent algorithms or language models to train their machines are free from “bias” of any kind, they said.

The amendments are likely to be notified soon by the Ministry of Electronics and Information Technology and mandate that such AI algorithms or language models that have inherent biases or are trained on such datasets which give out biased answers be barred from launching for open use.

ETtech

Apart from the safeguards for preventing bias in AI and language models, the IT Rules amendments may also bring new rules for deepfake, synthetic content and explicit instructions for platforms on loan apps, said one of the persons.

“The parameters for bias that we have right now are centred around caste, religion, community and national security, among other parameters. There will be a legal obligation on platforms to ensure all these parameters are followed, thereby expanding the ambit of trust and safety of the internet,” a senior government official told ET on condition of anonymity.

Discover the stories of your interest

The official said the government will conduct consultations on additional parameters to eliminate bias apart from those already known and understood to ensure safety.

These AI models, another official said, must undergo sandbox and stress testing to ensure that such biases are not present before releasing the tool in the market.

ET had reported on September 29 that the government is likely to introduce in the upcoming Digital India Bill similar provisions that seek to empower users to seek algorithmic accountability from internet intermediaries and social media companies.

The Digital India Bill, however, may now be released only after the general election, according to people in the know. They said the government’s decision to amend the IT Act may be prompted by the fact that the Digital India Bill is still some time away and developments around generative AI are progressing rapidly.

Also read | Generative AI’s wild 2023: how the viral technology took over the world

“The advisory on deepfake and synthetic content will be very specific on the dos and don’ts for platforms and how to tackle the issue. For the loan apps, we want these intermediaries to take the responsibility of preventing any kind of fraudulent advertising or stop such apps which do not have the requisite approvals from being hosted altogether,” said the official cited earlier.

Since the IT ministry is the “custodian of safety and trust” on the internet, its approach to preventing such loan apps from operating in India is likely to be either through a white-listing of the legitimate applications or placing a banking gateway which prevents the transactions through such apps or people, the official said.

“An online betting or fraudulent loan app can only function if they are allowed to connect to a legitimate banking solution. If we ensure that such people who operate these loan apps are denied access altogether to the banking system by the use of predictive algorithms, a lot of these platforms will wither away,” the official said.