Incident reporting is coming soon to an AI platform near you

India is framing guidelines for companies, developers and public institutions to report artificial intelligence-related incidents as the government seeks to create a database to understand and manage the risks AI poses to critical infrastructure.

- Jatin Grover

The proposed standard aims to record and classify problems such as AI system failures, unexpected results, or harmful effects of automated decisions, according to a new draft from the Telecommunications Engineering Centre (TEC). Mint has reviewed the document released by the technical arm of the Department of Telecommunications (DoT).

The guidelines will ask stakeholders to report events such as telecom network outages, power grid failures, security breaches, and AI mismanagement, and document their impact, according to the draft.

"Consultations with stakeholders are going on pertaining to the draft standard to document such AI-related incidents. TEC's focus is primarily on the telecom and other critical digital infrastructure sectors such as energy and power," said a government official, speaking on the condition of anonymity.

"However, once a standard to record such incidents is framed, it can be used interoperably in other sectors as AI is being used everywhere."

The plan is to create a central repository and pitch the standard globally to the United Nations' International Telecommunication Union, the official said.