Keynote by Mathias Cormann OECD Secretary-General India AI Impact
20 Feb 2026 15:00h - 16:00h
Keynote by Mathias Cormann OECD Secretary-General India AI Impact
Summary
At the India AI Impact Summit, the OECD thanked India and pledged evidence-based AI policy support worldwide [1-4]. It noted that AI could raise productivity by up to one percent annually, backed by nearly three-quarters of a trillion dollars in infrastructure spending [8-10]. Effective public policy, which previously enabled the internet and semiconductors, remains essential for AI benefits [11-12]. The OECD tracks AI compute capacity and venture capital, finding 61 % of AI VC now targets AI firms, 75 % of which is U.S.-based [15-17]. A report on generative AI agents shows half of developers plan to use them, urging better security and privacy [19]. AI incident reports rose from 92 to 324 per month between 2022 and 2025, leading to a common reporting framework [21-22]. The OECD released an AI Index for benchmarking and will launch a toolkit sharing global best practices [24-25]. Its Global Partnership on AI added Malta and Saudi Arabia, bringing membership to 46 countries [27-28]. For firms, it offers the Hiroshima AI Process Code, SME updates, and new due-diligence guidance [30-32]. About 27 % of jobs face high automation risk; flexible training is needed, especially as only 23 % of low-literacy adults receive AI training [35-38]. In partnership with the ILO, the OECD issued an Equitable AI Transitions Playbook and introduced the next data-sovereignty panel [39-40][41-47].
Keypoints
– AI’s transformative economic potential – The OECD highlights that widespread AI adoption could raise labor productivity by up to one percent annually across OECD and G20 nations, driving greater efficiency, lower costs, and higher living standards, while noting massive private-sector investment of “almost three quarters of a trillion dollars” in AI infrastructure this year [9-10].
– OECD’s data-driven support for policymakers – The organization provides evidence-based analysis on AI trends, including tracking global AI compute capacity, venture-capital flows (61 % of VC dollars now go to AI firms, with the U.S. receiving 75 % of deal value), and publishing reports on AI-agent usage and associated security, privacy, and accuracy needs [14-19][16-18].
– Monitoring and managing AI-related risks – The OECD collects and classifies AI incident data, showing a rise from 92 to 324 reported incidents per month between 2022-2025, and promotes a common reporting framework to ensure consistency and interoperability [20-22].
– Benchmarking and sharing best practices – New tools such as the OECD AI Index and an upcoming interactive toolkit will enable countries to assess progress against OECD recommendations and learn from international good-practice repositories [23-25].
– Addressing workforce impacts and equitable transition – The speech acknowledges that about 27 % of jobs are at high automation risk and that low-literacy adults have markedly lower AI-training participation (23 % vs. 61 %). It calls for flexible, modular training and introduces the “Equitable AI Transitions Playbook” developed with the ILO to guide up-skilling and reskilling policies [34-40].
Overall purpose/goal
The discussion serves to showcase the OECD’s role in guiding global AI policy-providing data, risk-monitoring, benchmarking tools, and international coordination-to harness AI’s economic benefits while mitigating associated risks, especially for workers and societies.
Overall tone
The tone is initially celebratory and forward-looking, emphasizing AI’s promise and the scale of investment. It then shifts to an analytical, evidence-based stance as Cormann presents data and policy tools. Mid-speech, the tone becomes cautiously concerned when addressing job displacement and training gaps, before concluding with a collaborative, solution-oriented call to action for governments, industry, and labor groups. The progression moves from optimism to measured caution and ends on a constructive, cooperative note.
Speakers
– Mathias Cormann
– Role/Title: Secretary-General, Organisation for Economic Co-operation and Development (OECD)
– Area of Expertise: AI policy, economic impact of artificial intelligence, public policy
– [S1]
– Speaker 2
– Role/Title: Moderator/Chair for the data-sovereignty panel at the India AI Impact Summit
– Area of Expertise: Event moderation, AI governance (implied)
– [S2]
Additional speakers:
– Sunil Gupta – Managing Director and Chief Executive Officer, Yota Data Services
– Nisubo Ongama – Chief Operating Officer, Kala
– Kala Sonia Vaigando – Founders Associate, Kala Limited
– Ms. Seema Ambasta – Chief Executive Officer, L & T, Vioma
– Mr. Orgo Sengupta – Founder and Research Director, WIDI Center for Legal Policy
The OECD Secretary-General opened the India AI Impact Summit by thanking India for its leadership in convening the global AI community after successful meetings in the United Kingdom, Korea and France, and reaffirmed the Organisation’s commitment to provide evidence-based analysis and policy guidance that supports responsible AI innovation worldwide [1-4].
He then highlighted the transformative economic promise of artificial intelligence. The OECD estimates that, with strong adoption, AI could raise labour productivity by up to one percent each year across OECD and G20 economies over the next decade – a boost that would translate into greater efficiency, lower costs and higher living standards. This optimism is underpinned by massive private-sector investment: big-tech firms alone plan to spend almost three-quarters of a trillion dollars on AI infrastructure in the current year [9-10].
Cormann stressed that such benefits will not materialise without effective public policy. He reminded the audience that the foundational technologies enabling today’s AI revolution – from internet connectivity to semiconductor supply chains – were themselves the result of deliberate policy interventions, and that similar pro-innovation, pro-adoption and pro-safety AI policies are now essential for AI [11-13].
To help governments design those policies, the OECD offers a data-driven service that maps the evolving AI ecosystem. It monitors the global distribution of public AI compute capacity to inform industrial-strategy decisions and supply-chain security, and tracks venture-capital flows, noting that 61 % of worldwide VC dollars (about US$259 billion) now target AI firms, up from 30 % three years ago, with the United States receiving 75 % of that deal value. A new report on the Argentic AI landscape, published last week, highlights that half of surveyed developers intend to use AI agents in their work, while flagging the need for improvements in security, privacy and accuracy [14-19][16-18][S46].
Risk management is another pillar of the OECD’s work. Its AI incident database records a sharp rise in reported hazards – from an average of 92 incidents per month in 2022 to 324 per month in 2025 – prompting the creation of a common framework for AI-incident reporting that promotes global consistency and interoperability [20-22][S46].
For benchmarking, the Organisation released the OECD AI Index, an evidence-based tool that lets countries assess progress against the OECD AI Recommendations, and announced an interactive toolkit to be launched later in the year, which will host a repository of good-practice case studies to facilitate peer learning [23-25][S51].
International coordination is pursued through the Global Partnership on AI (GPA). The partnership, designed to promote the responsible development and use of artificial intelligence grounded in the OECD’s landmarks, welcomed Malta and Saudi Arabia as its newest members, bringing total membership to 46 countries across six continents. The G-PI Council will reconvene later this morning to discuss next steps in the partnership’s work [26-28].
Beyond governments, the OECD supports businesses. It maintains the Hiroshima II Process Code of Conduct, introduced at the OECD II Action Summit in Paris last year, to foster transparency and accountability, and is now updating the framework to make it accessible to small and medium-sized enterprises. In addition, a new due-diligence guidance for responsible AI was published to help firms navigate the expanding landscape of regulations and voluntary standards [30-32].
Addressing the human dimension, Cormann noted that roughly 27 % of employment is in occupations at the highest risk of automation, underscoring the urgency of upskilling and reskilling programmes. His analysis revealed a stark disparity in AI-training participation: only 23 % of adults with low literacy engage in relevant training compared with 61 % of higher-literacy adults. He advocated for learning models that are flexible, modular and tailored to individual job contexts [34-38][S53].
In partnership with the International Labour Organisation, the OECD has produced the “Equitable AI Transitions Playbook”, which offers concrete policy examples for updating skills frameworks and launching up-skilling and reskilling initiatives that aim to ensure an inclusive AI transition while maximising the technology’s benefits and mitigating its disruptions [39-40][S54].
He then handed the floor to the next speaker. The second speaker thanked the Secretary-General for his insights, introduced the forthcoming data-sovereignty panel, and listed the distinguished panelists – Sunil Gupta (Yota Data Services), Nisubo Ongama (COO, Kala), Sonia Vaigando (Founders Associate, Kala Limited) and Ms. Seema Ambasta (Chief Executive Officer, L & T, Vioma) – before inviting the dignitaries to the stage, signalling a shift from high-level policy framing to a focused discussion on cross-border data governance [41-47].
Overall, the summit’s opening underscored the OECD’s role as a central, evidence-based hub that supplies data, risk-monitoring tools and benchmarking resources, while championing multistakeholder cooperation and inclusive policy design. By coupling quantitative forecasts of AI-driven productivity gains with concrete mechanisms for risk mitigation, corporate guidance and workforce upskilling, the Organisation aims to harness AI’s transformative potential responsibly and equitably, setting the stage for the detailed deliberations that follow.
India AI Impact Summit. And thank you to India for your leadership in bringing together the global AI community following the successful summits in the United Kingdom, Korea, and France. The OECD is proud to work with you and support policymakers, people, and businesses all around the world in harnessing the benefits of AI. And we do so with our unique data, evidence -based analysis, and policy guidance, aiming to promote responsible innovation and adoption while managing the potential risks along the way. In yesterday’s discussions, we heard about the wide -reaching potential impacts of AI development on our economies and societies. And of course, they continue to evolve as adoption accelerates and new applications are introduced. But one thing is clear.
These impacts are already a transforming and will become more transformative going forward. At the OECD, we estimate that with a strong level of adoption, AI could boost labor productivity by up to one percentage point every year across OECD and G20 countries over the next decade. Greater efficiency, lower costs, higher living standards, and the opportunities are also reflected in the scale of investment in AI infrastructure with almost three quarters of a trillion dollars in investment planned by big tech companies this year alone. Amid the rapid technological change and the massive investment flows, effective public policy is essential to allow AI to reach its full potential. Indeed, the foundational technologies that made this technological revolution possible were very much shaped and supported by public policy, from internet connectivity to semiconductor, supply chains, and everything in between.
Today, the OECD helps policymakers develop pro -innovation, pro -adoption, and pro -safety AI policies, drawing on the lessons of these previous interventions, sharing experiences at the cutting edge of AI policy, and identifying policy best practice. First, the OECD helps policymakers understand how AI technologies and business models are evolving and who the key players are in the AI ecosystem. We are tracking the global distribution of public AI compute capacity to help countries design their industrial strategies and assess opportunities to enhance AI supply chain security. We are also tracking global AI investment, with our analysis released earlier this week showing that 61 % of all venture capital investment worldwide, or $259 billion US, now goes to AI firms, which is up from just 30 % three years ago.
We are tracking the global distribution of public AI compute capacity to help countries Firms in the United States attract the largest share of venture capital by a wide margin, comprising 75 % of global I .I. venture capital deal value. Our analysis is also helping policymakers keep up with the latest technological developments. Our new report on the Argentic I .I. landscape, published last week, highlights that half of developers in recent surveys plan to use I .I. agents in their work, while identifying the need for progress on security, privacy and accuracy of I .I. agents to support further adoption. Second, we help policymakers track and classify I .I.-related risks. Our data on I .I. incidents shows that between 2022 and 2025, in just three years, the number of I .I.
incidents and hazards reported by the media increased dramatically, from 92 to 324 per month on average. The OECD common framework for reporting IA incidents helps promote global consistency and interoperability in IA incident reporting. And thirdly, we help policymakers benchmark their IA policies relative to their peers and international standards. Just yesterday, we released the OECD IA Index, which provides policymakers with an evidence -based tool to assess their progress in implementing the OECD recommendation on IA. We will also launch an interactive toolkit this year, which will feature a repository of good practices from around the world to support evidence -based peer learning. Fourth, we help governments coordinate their efforts internationally. Our integrated global partnership on IA was designed to promote the responsible development and use of artificial intelligence grounded in the OECD’s landmarks.
Thank you. G -PI, the G -PI Council, which we meet later this morning, to officially welcome our two newest members, Malta and Saudi Arabia, bringing G -PI’s membership to 46 countries across six continents. Beyond governments, we also provide analysis and recommendations to support II adoption of companies. The reporting framework for the Hiroshima II Process Code of Conduct launched at the OECD II Action Summit in Paris last year promotes transparency and accountability for responsible II innovation. We’re now updating that framework to support adoption by small and medium -sized enterprises. And yesterday, we published the OECD due diligence guidance for responsible II, which supports companies around the world in navigating a growing landscape of rules, regulations, and voluntary frameworks.
And we support people by providing recommendations for governments, business, labor, and other stakeholders to work together and to ensure everyone has the best possible opportunity to participate in and benefit from AI technologies. While AI adoption offers many exciting opportunities, it also carries the risk of job displacement for some. We estimate that taking the effects of AI into account, about 27 % of employment is in occupations that are at the highest risk of automation. It will be particularly important to ensure access to training opportunities for those who need the most. And on that front, our analysis shows that among adults with low literacy skills, only 23 % participate in relevant AI training, compared with 61 % of adults with higher literacy skills.
To improve participation in AI training among adults, learning needs to be more flexible, modular, and targeted to individual circumstances and job experiences. For this summit, together with the International Labour Organization, we have developed the Equitable AI. AI Transitions Playbook. which provides examples of policies to update skills frameworks as well as initiatives to upskill and reskill workers for an equitable II transition in closing to fully harness the enormous benefit and benefits and opportunities flowing from II while mitigating and managing some of the associated risks and disruptions we need to ensure governments industry labor and experts work together to support responsible adoption the OECD will continue to support this cooperation guided by our II principles so that II
Thank you so much, Secretary General of OECD. These remarks, we’re very grateful for your remarks. For the next panel on data sovereignty, we have Mr. Sunil Gupta, Managing Director and Chief Executive Officer, Yota Data Services. We have Nisubo Ongama, COO, Kala Sonia Vaigando, Founders Associate, Kala Limited. We have Ms. Seema Ambasta, Chief Executive Officer, L &T, Vioma. And this session is being moderated by Mr. Orgo Sengupta, Founder and Research Director, WIDI Center for Legal Policy. May I request all the dignitaries to come up on stage, please.
“The OECD estimates that, with strong adoption, AI could raise labour productivity by up to one percent each year across OECD and G20 economies over the next decade.”
While the report gives a specific 1 % annual productivity boost, the knowledge base cites a PwC study showing AI-intensive industries experiencing productivity growth nearly five times faster than average, and an IDC forecast that AI could add $19.9 trillion to the global economy by 2030, providing broader context on AI’s economic impact [S42] and [S62].
“Big‑tech firms alone plan to spend almost three‑quarters of a trillion dollars on AI infrastructure in the current year.”
The knowledge base notes that major technology companies are significantly increasing capital expenditures for AI data centres and that overall AI-related spend could approach $2 trillion over the next 5-10 years, but it does not give the exact $0.75 trillion figure cited in the report [S63] and [S66].
“61 % of worldwide VC dollars (about US$259 billion) now target AI firms, up from 30 % three years ago, with the United States receiving 75 % of that deal value.”
Industry observations in the knowledge base describe a shift toward leaner AI start-ups and heightened venture-capital activity, but they do not provide the precise 61 % share or the $259 billion amount; the trend is corroborated in a discussion of VC dynamics in AI startups [S78] and big-tech investment patterns [S63].
“The Global Partnership on AI (GPA) welcomed Malta and Saudi Arabia as its newest members, bringing total membership …”
The knowledge base mentions the GPA’s co-chairmanship by Korea and Singapore and the existence of the OECD AI Policy Observatory, confirming the partnership’s structure, but it does not list Malta or Saudi Arabia as recent members [S69].
“A new report on the Argentic AI landscape highlights that half of surveyed developers intend to use AI agents in their work.”
Separate analysis of AI agents shows rapid growth in interest and adoption, indicating that a sizable proportion of developers are exploring agent-based tools, which aligns with the report’s finding though the exact 50 % figure is not specified in the knowledge base [S65].
The two speakers show clear convergence on two fronts: (1) mutual appreciation of the OECD’s evidence‑based, policy‑support role in AI, and (2) a shared belief that international, multistakeholder collaboration is essential for responsible AI and data governance. Beyond these, an unanticipated overlap emerges around the theme of inclusivity—both in AI skills development and in the composition of forthcoming discussion panels.
Moderate consensus. The agreement is limited to high‑level acknowledgments of the OECD’s role and the need for cooperation, which bodes well for coordinated policy action but leaves substantive policy details (e.g., specific AI risk mitigation measures or training program designs) unaddressed in the short exchange.
The transcript contains only a keynote by Mathias Cormann and a brief hand‑over by Speaker 2. No opposing viewpoints or substantive debates are presented; the two speakers are aligned in acknowledging the OECD’s contributions and in moving the agenda forward.
Minimal – the interaction is essentially cooperative, implying that any policy disagreements are not evident in this segment and are unlikely to impede the broader discussion on AI and data governance.
The discussion was driven by a series of strategically placed insights from Mathias Cormann that alternated between highlighting AI’s massive economic promise and exposing its emerging risks and inequities. Each pivot—productivity gains, capital flows, incident spikes, job‑displacement figures, and training gaps—served as a turning point that broadened the conversation from pure optimism to a nuanced, policy‑centric dialogue. By coupling hard data with concrete policy tools (the OECD AI Index, incident‑reporting framework, and the Equitable AI Transitions Playbook), the speaker transformed abstract trends into actionable agendas, setting the stage for the subsequent panel on data sovereignty and signaling a shift from analysis to implementation.
Disclaimer: This is not an official session record. DiploAI generates these resources from audiovisual recordings, and they are presented as-is, including potential errors. Due to logistical challenges, such as discrepancies in audio/video or transcripts, names may be misspelled. We strive for accuracy to the best of our ability.
Related event
