Is India prepared to tackle 369 million cybersecurity incidents and AI challenges ahead?

According to Data Security Council of India's 'India Cyber Threat Report 2025', 369.01 mn security incidents were detected across 8.44 mn endpoints throughout 2024. This translates into 702 potential security threats on an average every minute. Between 2021 and end-2024, detection of behavioural-based cyber threats jumped from 5 mn to 54 mn. Healthcare (21.82%), hospitality (19.57%) and BFSI (17.38%) are the top targeted sectors.

While there have been advancements in detection capabilities, gaps in India's cybersecurity infrastructure and 42% talent shortage of skilled manpower in the sector should be fixed on priority. AI-enabled deepfakes are being increasingly used to bypass traditional efforts to combat money-laundering, terrorism financing, proliferation financing controls, and manipulation of systems like KYC, customer due diligence (CDD) and digital ID verification.

These challenges highlight the need to move cautiously in AI adoption in our processes and work. The UAE comes out on top of Microsoft's 'Global AI Diffusion Report 2026', with an adoption rate of 73.3%, whereas India ranks 64th with a diffusion rate of 18.5%. However, the leading AI innovator countries, the US and China, rank 21st (33%) and 67th (17.5%) respectively, which reflects their cautious approach in AI adoption. When it comes to global regulation to control the challenges posed by AI, especially agentic AI, the US and China have shown little inclination to develop global governance and compliance standards.

Adequate guard rails and framework should be in place to accredit AI models, so that their safety, security and accuracy can be verified and algorithmic liability can be fixed. GoI has already issued 'AI Governance Guidelines' in Nov 2025. An AI Safety Institute (Aisi) has also been created under IndiaAI Mission, which provides technical expertise to carry out research, perform safety testing, and develop standards to address potential risks from technology.

Best practices are being followed by some countries, including Britain, where the AI Security Institute employs 'universal jailbreaks' technique to test and verify robustness of AI models. The EU has even enacted a legislative and administrative framework in 'Artificial Intelligence Act', which classifies AI according to its risk - unacceptable risk is prohibited, high-risk AI systems are regulated, limited-risk AI systems are subject to lighter transparency obligations, and minimal risk is unregulated - with majority of obligations falling on providers (developers) of high-risk AI systems. Brazil, too, is actively developing an AI Bill to regulate the tech.

India, too, needs a comprehensive National AI Act, which clearly defines the boundaries for the tech, sets compliance standards and fixes algorithmic liability. GoI has already signalled that the time is now ripe to have a separate national-level AI legislation.

MeitY's Aisi and Centres of Excellence (CoEs) need to be further strengthened and empowered to access all AI models, applications and agents being used in the country, so that the safety, accuracy and security of these models can be tested and verified for large-scale use by individuals and organisations.

Even though the overall rate of AI adoption in India is low, it has witnessed rising active users of ChatGPT, Gemini, Copilot and DeepSeek (globally ranked 3rd), and revenue of Anthropic has doubled in last few months. Since private sector in India leads in adoption of AI, unregulated, easy and free access to AI models - even unaccredited and unvalidated ones - has posed serious challenges in terms of data protection, safety of confidential information and cybersecurity.

A 2025 global KPMG-University of Melbourne study reveals 57% of employees admitted hiding their use of AI at work. Accordingly, organisational silence on use of AI by employees has made the challenges even more complex. Absence of global and national regulatory legal framework fixing algorithmic liability as sequel to misuse or malfunction of AI agents or apps has become quite a challenge.

All organisations and businesses should develop an organisational protocol for adoption of AI, and actively participate in government initiatives to put in place a robust legal and administrative framework to tackle these challenges.