India AI Impact Summit 2026 – Day 1
- Sarvada Vartalap
- Feb 16
- 9 min read
Updated: Feb 25
The India AI Impact Summit 2026 in New Delhi marked a quiet yet consequential inflection point. For the first time, a global AI summit shifted to the Global South, signalling not just a geographic move but a rebalancing of participation, scrutiny, and imagination about AI’s future.
Where earlier summits in the U.K., South Korea, and France were shaped by frontier-model debates and regulatory architectures typical of highly digitised economies, New Delhi framed its contribution around progress anchored in inclusion.
The scale was striking: an estimated 250,000 participants. For context, India’s largest cricket stadium seats about 100,000. Queues outside Bharat Mandapam resembled match-day crowds—an apt signal of how deeply AI has entered public consciousness.
Day 1 met that anticipation with discussions that were technical yet contextual, balanced, and often provocatively constructive.
Walking through Bharat Mandapam—from West Wing to East Wing and across levels L1 to L15— traversing through topics which are of paramount significance: safety, inclusion, governance, innovation.
In cricket context, lets do a 300 word highlights! A unifying theme across sessions was the primacy of trust as the foundational currency for AI adoption. Whether in cybersecurity, standards-setting, or AI governance, one constant theme emerged trust cannot be retrofitted— it must be engineered into products, data flows, and institutional processes from inception. The importance of context-sensitive design was highlighted, noting that different intermediaries carry distinct duties of care and accountability.
Competition emerged as another critical dimension. India’s digital markets remain dominated by a handful of large platforms, fuelled by data concentration and high entry barriers. Open-source AI was positioned as a key lever for democratising innovation and strengthening India's domestic AI ecosystem. Complementing this, discussions on India’s frugal innovation strategy reaffirmed the nation’s dual focus: developing efficient small models while simultaneously investing in sovereign foundational models to maintain global competitiveness.
Agentic AI—where humans define intent and AI agents execute actions—showcased both promise and risk. Delegates emphasised the need for explicit consent, verifiable agent identity, and transparent chains of custody. The paradigm has significant potential for MSME empowerment but requires strong global interoperability and trust frameworks.
In the media domain, AI’s role in reshaping information flows prompted an urgent call for accountability, transparency, and recognition of journalism as a public good. A nine-point roadmap highlighted fair value for journalistic content, mandatory attribution, and penalties for hallucinations. As AI enters democratic spaces, higher standards of responsibility must guide its design and deployment.
On sovereignty, speakers underscored that compute, data, and chip control define geopolitical power. India’s dependence on global monopolies in design, manufacturing, and export controls places strategic limits on innovation. Policymakers were urged to focus less on technical minutiae and more on outcomes: economic value creation, equity impacts, and societal risk.
A recurring caution was that legislation alone is insufficient—enforcement determines impact. Simultaneously, India’s outdated copyright framework requires urgent clarity on AI training exceptions to avoid stifling innovation while safeguarding creators’ rights.
Overall, Day 1 showcased India’s emergence as a serious contributor shaping global AI governance, innovation pathways, and norms of responsible deployment. The key message was unmistakable: India’s AI future must be rooted in trust, interoperability, sovereign capability, and inclusive design.
Now, for detailed insights of Day 1, please scroll below.
Embedding Trust in Innovation
Trust is emerging as the defining currency of the AI era: no sector can scale sustainably without robust, reliable infrastructure. From cybersecurity to global standards to the future of AI testing, one constant message stood out—trust is not a by-product of innovation; it is a precondition for growth.
As AI advances, policymakers and academics converged on a pragmatic approach: prioritise clear, practical guidance over sweeping new laws. The goal is transparency and scaled adoption through predictable, implementable rules.
Structured AI assessments are becoming essential to validate compliance with existing regulations and to determine whether underlying models are reliable, practical, and fit for sector-specific deployment. For many organisations, regulation is evolving from a narrow compliance obligation into a strategic governance lever—and a source of competitive advantage.
AI governance cannot be built in isolation. Cybersecurity teams, standards bodies, certification authorities, and technology companies must collaborate across the lifecycle to define how trustworthy systems are designed, tested, deployed, and monitored.
Critically, trust must be embedded at design time. Compliance, safety, and transparency cannot be retrofitted; they must be intrinsic to system architecture and development practices. Equally, trust-by-design must account for the role of intermediaries: social, commercial, and informational intermediaries carry distinct duties of care and accountability profiles. Trust mechanisms should therefore be contextual, not one-size-fits-all. As systems grow in complexity and unpredictability, rigorous evaluation is moving to the centre of AI governance. Strong testing frameworks are no longer optional; they are the primary mechanism for responsible, transparent deployment.
The emerging consensus points to a future where:
AI risks are proactively identified and mitigated;
Systems remain explainable, auditable, and continuously monitored;
Trust-building is integrated into the earliest stages of design; and
Global standards and cross-border interoperability enable responsible scale.
Ultimately, trust is not aspirational—it is the structural backbone that will determine whether AI scales safely, equitably, and sustainably.
AI and Cybersecurity : Rising Risks, Smarter Responses
AI-assisted workflows are materially improving incident detection and response. Yet the same technologies have accelerated the risk of cybercrime, enabling even novice attackers to generate malware, phishing content, and other threats. To embed trust in this environment, four priorities were underscored:
Calibrate autonomy based on cyber-risk assessments;
Build secure-by-design AI applications;
Ensure explainability to support audits and investigations; and
Increase supply-chain transparency—AI systems cannot remain black boxes.
CERT-In is collaborating with global authorities on transparency guidelines and a risk-based developer checklist aligned with international standards (including ISO frameworks).
The Standards Pipeline : Building Reliable and Validated AI
Trusted AI depends on standards that are reliable, traceable, and verifiable. While industry has embraced quality and risk-management frameworks (e.g., ISO 14001), AI demands an expanded suite of testing and evaluation protocols—including benchmarks for hallucination rates and other governance-critical behaviours.
Two priority gaps surfaced:
Heavy reliance on synthetic data in the startup ecosystem—necessitating scalable, shared data platforms; and
Predominantly English-centric global standards and models—reinforcing the need for robust, bias-resistant Indic language models.
A recurring analogy drew from UPI’s open architecture: interoperable data frameworks could enable AI models to scale with trust, inclusivity, and consistency.
The Path Forward : Trust as a Shared Responsibility
AI governance cannot be built in silos. Cybersecurity teams, standards organisations, certification authorities, and technology companies must act in concert. The shared vision is one where risks are proactively managed, systems are explainable and auditable, and global alignment on standards accelerates innovation. In this era, trust is the foundation on which all responsible AI progress rests.
Competing to Innovate : How Competition Accelerates AI Innovation
Digital markets are increasingly concentrated, with competitive dynamics skewed toward a few dominant players—an experience India has seen in app-store and operating system markets. Control over data ecosystems compounds market power, while the fast-evolving digital economy demands continual innovation and significant capital—barriers that make it hard for smaller firms to enter, compete, or scale.
For startups, matching the service quality, technological sophistication, and reach of incumbents remains challenging. User trust, network effects, and habitual reliance reinforce incumbency. Against this backdrop, open-source AI can rebalance dynamics by lowering entry barriers and expanding access to advanced capabilities, thereby stimulating innovation and competition.
India is shaping a strategic path that welcomes global investment and innovation while strengthening domestic capacity. An emphasis on openness, interoperability, and user choice will be vital to building a healthy, competitive, and resilient marketplace.
From AI User to Creator : The next leap of India’s AI Innovation
India’s AI strategy is anchored in frugal innovation—maximising societal impact with optimised resources. Rather than focusing exclusively on massive large language models, India is prioritising smaller, scalable, and efficient models that can be rapidly deployed and adapted across contexts, ensuring accessibility and inclusion.
Frugality does not imply a ceiling on ambition. To remain globally competitive, India must also invest in large-scale foundational models that compound value, accelerate downstream innovation, and serve as core infrastructure for future applications.
A central insight was that AI trust cannot be imported; it must be engineered. Externally built models reduce visibility into training data, evaluation, and optimisation, creating trust deficits. India must build and govern its own systems, embedding compliance, safety, and reliability across the lifecycle—through compliance by design, clear and quantifiable trust metrics, and supply-chain visibility. For India, trusted AI is strategic, not optional.
Mr. Jitin Prasada, Union Minister of State for Electronics & Information Technology, underscored the need for collective action across government, industry, and academia. He highlighted an agile policy stance grounded in active stakeholder engagement. Key enablers included reducing GPU costs, expanding AI skilling, favouring innovation over over-regulation, democratising access, scaling solutions, and leveraging India’s talent base.
Agentic Commerce : Trust, tokens & know your agent for AI Economy
A fundamental shift is underway: humans increasingly articulate intent and outcomes, while AI agents execute complex, multi-step actions—from discovery and decision-making to transactions. Users now orchestrate results rather than perform every step. Human intent and explicit consent must remain paramount for responsible deployment.
Real-world purchasing is contextual and layered. Where decision logic is nuanced or implicit, agentic AI faces difficulty; by contrast, rule-bound processes (e.g., company registration, compliance, administrative workflows) are more amenable to agent-based systems.
Trust emerged as the decisive factor for adoption, with practices such as:
Multi-stage, explicit consent before transactions;
Strictly limited, purpose-bound data access;
Intentional human-in-the-loop oversight even when full automation is viable; and
Automatic deletion of data on task completion.
For responsible scale, agents must be verifiable, accountable, and traceable, including:
Cryptographically verifiable agent identities;
Clear chains of custody for agent-initiated actions; and
Robust grievance-redressal and escalation mechanisms.
Merchants are advocating agent adoption not only to streamline transactions but to gain richer, unstructured insight into consumer behaviour. Unlike clickstream data, agent interactions reveal reasoning, questions, comparisons, and trade-offs—driving demand for interoperable agent platforms that operate across ecosystems.
Affordable, interoperable agentic infrastructure could materially benefit MSMEs by lowering entry barriers and operating costs. Given the cross-border nature of agents, global governance, interoperability standards, and shared trust frameworks will be essential to safeguard equity, inclusion, and responsibility.
AI & Media : Opportunity, Responsibility & Road Ahead
AI is transforming the information lifecycle—from gathering and curation to distribution, trust, and action. As information becomes commoditised, trust grows scarcer and more valuable. Journalism, unlike general content creation, directly shapes public opinion and democratic discourse; trust therefore arises from institutions that practise editorial judgment, verification, and accountability—not from AI.
The "human moat"—a deliberate human role at critical decision points—was framed as an "Accountability Sandwich": human intent initiates, AI assists processing, and human judgment delivers the final decision.
As AI systems enter democratic and public conversation, they must meet higher responsibility standards. The EU AI Act’s mandatory labelling was cited as evidence that transparency is becoming a baseline expectation.
A nine-point agenda was discussed:
Fair value for journalistic content used in AI, with transparency on model ingestion;
mandatory attribution and traceability—labelling cannot be optional;
formal recognition of journalism as a public good;
rewards for stories with measurable social impact;
economic value for verified reporting;
strict penalties for AI hallucinations;
end asymmetric rules for legacy media versus social platforms;
treat public attention as the scarcest resource; and
insist on reciprocity—what technology firms give back for attention and data.
An unresolved question persists: Who bears responsibility for harmful or incorrect content—the creator, distributor, or platform? Clear accountability frameworks are essential; consumers also have agency in demanding labelling, accuracy, and transparency.
Implementation priorities included:
Mandatory labelling of original, verifiable content by governments and major platforms;
ensuring access to high-quality datasets for Indian AI models;
building government-led infrastructure for trustworthy AI and resilient media; and
expanding public funding and policy support, including state-backed training on professional journalism (as in Norway and South Africa).
Power, Protection & Progress : Legislating for the AI Era
AI today is about compute, data, and control. Nations that command chip design, fabrication, and large-scale compute infrastructure will shape 21st-century economic power. AI policy is thus a question of competitiveness and sovereignty, especially for India.
AI is an amplifier: in democracies it can deepen participation and governance; in unequal systems it can widen divides. Without broad, affordable access and safeguards, it risks eroding public trust.
India’s ecosystem faces three global monopolies:
Design monopoly
Manufacturing monopoly
Export control monopoly
Consequently, compute is not merely a supply-chain concern but a sovereignty imperative. India’s talent is formidable, but without sovereign compute, talent translates to dependency, not leadership.
Policymakers need not master GPUs or neural architectures; they must grasp how AI drives value creation, who benefits from locally generated data, whether AI narrows or widens inequality, and how it affects children, democracy, and trust. Like nuclear regulation, AI policy should focus on outcomes and risk, not internal mechanics.
Legislation is only a starting point; enforcement determines impact. The record of major technology firms circumventing penalties and remedies—domestically and abroad—shows that without robust enforcement, regulation risks becoming symbolic.
Copyright law also needs modernisation. Ambiguity around fair dealing in the AI era leaves developers and creators in legal limbo. A balanced, contemporary fair-use framework can respect creators’ rights while enabling innovation—echoing Israel’s 2020 advisory permitting AI training on copyrighted data under fair-use principles.
On the 1st day of Digital AI Summit, Sarvada Legal attended:
Embedding Trust in Innovation: AI Governance and Quality Infrastructure for Growth;
Effective AI Assessments, Verification and Assurance: Establishing the Foundations for Responsible Confidence in AI;
Competing to Innovate: How Competition Accelerates AI Innovation;
High-level discussion on APAC Centre for AI;
From AI User to Creator: The next leap of India’s AI Innovation;
Agentic Commerce: Trust, tokens & know your agent for AI Economy;
AI & Media: Opportunity, Responsibility & Road Ahead;
Power, Protection & Progress: Legislating for the AI Era;
Culturally-Grounded AI: How Social Norms Can Inform AI Systems;
Empowering the Human Edge: Building a Future-Ready Workforce in the Age of AI’; and Responsible AI Hub: Responsible deployment and use of AI systems in Social Welfare Delivery’.
Comments