AI Model Testing: Vaishnaw Warns Against Untested AI Release

New Delhi, Mar 4 (PTI) Social media firms and internet companies need to take the onus of their work, Union Minister for Electronics and Information Technology Ashwini Vaishnaw said on Monday, asserting that no untested AI model can be made public as it is not good for society.

The minister also said that the government advisory on Artificial Intelligence (AI) is not binding but it apprises firms that they should make their AI models public only after testing.

Deployment of AI models should be done responsibly, Vaishnaw said, adding, "It is important for citizen's safety, democratic process. Therefore, an advisory has been issued that any AI model should be properly tested before being made available. It is applicable on social media platforms (as well)."

He was replying to a question on the government's AI advisory.

In the advisory issued to intermediaries and platforms on March 1, the Ministry of Electronics and Information Technology warned of criminal action in case of non-compliance.

"All platforms, intermediaries and enabling software shall be held accountable for any breaches and non-compliance with provisions would result in penal consequences," it said.

The advisory came within days of Google's AI platform Gemini making uncharitable comments about Prime Minister Modi's policies.

Without naming Google, the minister said, "Some people have apologised that they didn't test the model properly before uploading it. Social media has a wide reach. Social media platforms and owners of internet firms should take responsibility for what they are doing.