Centre issues transparency, safety guidelines for AI

CERT-In's guidelines mandate six-hour breach reporting and 180-day data logs, but remain non-binding for now. 

Shouvik Das
Published4 Nov 2026, 08:57 PM IST
At the core of the guidelines is a proposed classification system that would place AI models and applications in four tiers of criticality — low, moderate, high, and critical.
At the core of the guidelines is a proposed classification system that would place AI models and applications in four tiers of criticality — low, moderate, high, and critical.(REUTERS)

India's top cybersecurity agency has laid out a set of security, transparency and accountability requirements for artificial intelligence (AI) systems, seeking to establish minimum standards for companies building and deploying AI.

The 107-page guidelines, titled Artificial Intelligence Security Auditing Guidelines: Framework for Secure, Ethical and Auditable Artificial Intelligence Systems, was published on Wednesday by the Indian Computer Emergency Response Team (CERT-In).

While it applies to all AI developers and service providers, including US-based companies behind popular applications ChatGPT, Claude and Gemini, it is legally binding only for cyber security auditors that are empanelled with Cert-In. Put simply, while the guidelines are themselves not binding on companies building AI, they are the basis on which AI systems within companies will be audited. If found inadequate, and once India establishes its AI regulations, failing to comply with these guidelines could carry consequences,

Also Read | India bets on ‘sealed rooms’ to train AI on sensitive data without exposing it

The framework identifies about 10 areas in which companies building AI models, applications and related systems should seek to meet specified standards. Among its requirements, companies must report AI-related security breaches to CERT-In within six hours and maintain logs of AI systems' usage and operations for 180 days.

At the core of the guidelines is a proposed classification system that would place AI models and applications in four tiers of criticality — low, moderate, high, and critical. The classification would be based on eight identifiers, including the severity of potential impact and sensitivity of data.

The framework also envisages independent audits and security assessments. Experts said its risk classification and audit structure appear to draw on the European Union's (EU) approach. Unlike the EU AI Act, however, India's guidelines are not legally binding.

An official with direct knowledge of developments said the guidelines “are for auditors to use whenever they are examining the AI systems being used by any large company, for which a cyber security audit is crucial to pass.”

Discussions of whether the guideline should evolve into law are “a completely separate matter, and are progressing with the right industry stakeholders inside Meity (Ministry of Electronics and IT), but these guidelines remain firmly for security audits,” the official said.

“The standards are also up to global ISO standards, and given that AI is a rapidly evolving field, establish clear best practices that make AI systems in India more accountable,” the official added.

The guidelines argue that “AI governance cannot be addressed solely through ethical principles or conventional cybersecurity controls.”

Also Read | Apple's record India quarter has more to it than meets the eyes

“Effective assurance requires organizational accountability, technical testing, data and model governance, continuous monitoring, human oversight, incident response and third-party assurance. Accordingly, the framework provides auditable requirements proportionate to the nature, scale, autonomy and potential impact of each AI system.”

Global push

India joins a growing international effort to establish greater oversight of AI systems, whose algorithms have often been criticized for operating as black boxes.

The European Union has adopted a legally binding AI Act, while the US has developed voluntary AI standards through the National Institute of Standards and Technology (NIST). China requires generative AI services to undergo government security checks before launch.

Industry stakeholders said India's guidelines do not offer a novel approach, adding that their non-binding nature may mean that there’s no real compliance on the ground.

“If you look at the work that the EU, and Nist in the US have done, they’re pretty much in the same vein. India, in fact, has had similar state-level AI usability frameworks, such as Tamil Nadu’s Safe & Ethical AI Policy, 2020. This, therefore, is nothing new. Typically, non-binding guidelines generally do not have a major impact on any industry, and it remains to be seen if the guidelines’ framework is imbibed into law, which Meity has publicly acknowledged is being worked upon,” said Jibu Elias, head of responsible computing in India at open-source internet body, Mozilla Foundation.

Also Read | Call for new AI intermediary definition splits New Delhi

The latest framework is the second initiative aimed at AI governance and transparency to emerge in less than a week. On 30 October, independent think-tank iSPIRT, in partnership with MeitY, unveiled its Data Empowerment and Protection Architecture (DEPA) for AI.

Officials had told Mint at the time that the DEPA framework would also likely undergo industry consultations before the government considers whether to incorporate it into law.

About the Author

Shouvik has been tracking the rise and shifts of India’s technology ecosystem for over a decade, across print, broadcast and web-first platforms. He's been a tinkerer of machines and PCs since childhood, a habit he was thrilled to convert into his profession. This has led him to fascinating experiences of technologies around the world, which is what keeps him hooked to his job.<br><br>Shouvik likes to believe that he is one of the few technology journalists in India who can also code. He has also been writing about the rise of AI well before it became a household name, and has met some of the most fascinating people over the years through his work.<br><br>Shouvik writes about AI, Big Tech, data centres, electronics, semiconductors, cybersecurity, gaming, cryptocurrencies, and consumer technologies. He is most fond of the stories he has written during his time here at Mint, for which he also writes 'Transformer', a weekly technology newsletter, and hosts 'Techcetra', a weekly technology podcast.<br><br>Outside of work, Shouvik spends most of his time with Pixel, whom he believes is the world's best dog. He is also an avid reader, a toy collector, a gamer and a frequent traveller.

Get Latest real-time updates

Catch all the Business News, Market News, Breaking News Events and Latest News Updates on Live Mint. Download The Mint News App to get Daily Market Updates.

HomeAIArtificial IntelligenceCentre issues transparency, safety guidelines for AI
More