Any instances of bias in the content generated through algorithms, search engines or AI models of platforms like Google Bard, ChatGPT, and others will not be entitled to protection under the safe harbour clause of Section 79 of the Information Technology Act, minister of state for electronics and IT, Rajeev Chandrasekhar told Fe. Safe harbour provision gives social media platforms such as Twitter, Facebook, Instagram, etc, legal immunity against content shared by users on the platforms.
Disclaimers which several intermediaries put below such AI generated content that it’s experimental, will not suffice. “Intermediaries have no business to put out any experimental content to the public,” he said.
The minister said that the government is working on a Digital India Bill, which will address AI regulations in detail, but that does not mean that currently there’s a vacuum. “There are rules and guardrails in place which address misuse of AI, be it deepfakery or bias in search engines, algorithms, etc. The IT Rules apply to intermediaries and any violation by them on any count, including AI will be dealt with through these rules,” Chandrasekhar said.
However, the minister said that in a change of approach the government is now telling aggrieved parties to file an FIR, so that the courts can decide on the matter. In the process the government will also clarify its position. “Under the IT Rules if the government asks the intermediary concerned to take down any misleading or biased content, then people say that the move is against freedom of speech and expression. So, we are encouraging citizens to file FIR so that the matter can be dealt by the courts,” Chandrasekhar said.
Earlier in the day, through a post on social media site X, the minister had responded to a user that search bias, algorithmic bias and AI models with bias are real violations of the safety and trust obligations placed on platforms under Rule 3(1)(b) of IT Rules under regulatory framework in India. Those who are aggrieved by this can file FIRs against such platforms and safe harbour will not apply to these cases.
The user had flagged a screenshot of Google Bard which refused to summarise an article by a right wing online media on the ground that it spreads false information and is biased.
Chandrasekhar told Fe that he’s not endorsing the screenshot shared but merely informing the user about the law and the steps that can be taken by the user.
According to legal experts, intermediaries need to do much more thorough due diligence with AI generated content put out by their platforms because they will have no immunity here under the safe harbour clause. Safe harbour applies to content generated by third party which intermediaries take down upon being flagged as misleading or harmful.
“These generative AI platforms are trained on large language models (LLMs) and these training data is available on the Internet. Therefore, it is difficult to say that there will be a zero bias model,” said Jaspreet Bindra, founder and managing director of The Tech Wisperer, a technology advisory and consulting firm.
“AI biasness can lead to problems like perpetuating stereotypes, reinforcing harmful narratives, creating unequal representation, etc,” said Anupam Shukla, partner at Pioneer Legal. “No single technique can be used to fully mitigate bias. A combination of techniques is the best course of action. Models like statistical analysis, algorithms to undertake fairness evaluation, and human analysis can be used to ascertain bias in GenAI systems,” Shukla added.
