Scaling Trusted AI: How France and India Are Building Industrial & Innovation Bridges

Abstract

The session opened with Business France highlighting the breadth of the French delegation and announcing a series of strategic partnerships spanning AI, space, healthcare and industrial innovation. Julie Uge delivered a visionary keynote emphasizing the shared Franco‑Indian values of trustworthy, low‑impact AI and showcased French deep‑tech startups ready to partner with Indian ecosystems. A moderated high‑level panel then dissected what “trust” means across cloud, quantum, defense, industrial and enterprise AI, stressing end‑to‑end governance, explainability, security and the need for ecosystem‑wide mind‑set change. The programme concluded with a rapid‑fire keynote on multilingual AGI by Prof Raj Reddy and a second panel probing AI’s role as a scientific instrument, reproducibility, responsible governance and the potential for North‑South collaborations to accelerate climate‑resilient and health‑focused research.

Detailed Summary

• Business France opened by celebrating the presence of Prime Minister Modi and President Macron at the pavilion.
• The French AI delegation comprised ~100 companies covering Quantum‑Ready, Photonics, Secure Edge AI, Mobility Systems, Cyber‑Security, Digital Twin, Green‑Tech.
• Key partnership signatures announced:
  1. Dacia Technology & GT Solved – strategic partnership for engineering automation (signed at the French consulate, Bangalore).
  2. Exot Trail & Druva Space – contract to deliver 14 satellite propulsion systems.
  3. H Company & St James Hospital – collaboration to improve hospital efficiency and save lives.
  4. North France Invest & TIAAB – creates a bridge between a leading European industrial region and India’s innovation ecosystem.
• Emphasis on the role of Business France, La French Tech, Numium, Yuja Advisory, Artyos, and the Franco‑Thai Chamber of Commerce in orchestrating the event.

2. Keynote – French Tech Vision for India (Julie Uge)

Theme	Highlights
Scale of Talent	India graduates 1.5 million engineers annually – the world’s largest engineering pipeline.
AI Ecosystem Ranking	Paris now ranks among the top‑3 AI hubs globally (San Francisco, New York, Paris).
Trusted AI Values	“Trustworthy, low environmental footprint, positive impact for humanity.”
Concrete French Start‑ups	AgriCo (farm‑to‑market), WhitLab Genomics (AI‑accelerated gene‑therapy), Candela (scalable quantum tech), H Company (advanced AI agents).
Franco‑Indian Complementarity	France contributes deep‑tech excellence & industrial capacity; India contributes scale, 200 k startups, market size.
Cultural Learning	French pre‑scheduled meetings vs. Indian flexibility – both needed for successful collaboration.
Call‑to‑Action	“Build strategic partnerships, share Indian wisdom, accelerate joint R&D.”

3. High‑Level Panel – Scaling Trusted AI Across Sectors

Moderator: Arun Sadheesh (TNP Consultants)

3.1. Defining Trust – A Common Thread

• Arun Sadheesh opened by recalling PM Modi’s “manav” concept and President Macron’s UPI example – trust enables scale.
• Trust is positioned not as optional “bolt‑on” but as architectural, foundational to all AI layers.

3.2. Panelist Contributions

Speaker	Core Points on Trust
Nila Khan (Tata Communications)	Trust = “I’ve got your back, I won’t fail you.” Emphasised zero‑trust networking, data lineage, end‑to‑end governance across cloud‑edge pipelines; AI must be explainable, auditable, resilient.
Valerian Gies (Quantela/Candela)	Introduced five pillars: trustability (traceability), predictability, verifiability (benchmarking), security, accountability. Presented Merlin – a benchmarking framework for quantum‑AI to create reproducible results and shared baselines.
David Sadek (Thales)	Outlined four pillars of trustworthy AI: validity (formal safety proofs), security (adversarial resilience), explainability (human‑readable rationale), responsibility (ethics, carbon‑footprint, frugal AI). Shared concrete example: digital copilot must be able to justify a left‑turn decision.
Sandeep Kumar Saxena (HCL)	Trust begins with leadership; built AI‑driven internal growth‑engine (“voice‑first dashboards”, AI‑powered forecasting). Stressed iterative learning and patient, transparent rollout to earn user confidence. Highlighted AI solutions (fraud detection, compliance, skilling) built in‑India, for‑India, for the world.
Tanuj Mittal (Dassault Systèmes)	Trust evolved from accuracy‑only to data‑lineage, people‑in‑the‑loop, virtual‑twin simulations, and full‑life‑cycle governance (concept → decommission). Emphasised simulation of AI outcomes in real‑world Indian contexts to validate models before deployment.

3.3. Mind‑Set Shift & Ecosystem Thinking

• Neela Khan (representing Tata) called for an ecosystem mindset—partnering (e.g., with Thales) to keep the trust architecture intact across the value chain.
• Valerian Gies urged breaking walls between quantum and AI, fostering a community through Merlin for reproducibility.
• David Sadek highlighted that trust must be proved, not merely declared, especially for safety‑critical systems.
• Sandeep Saxena summed up the shift in a single word: “Adaptability.”

3.4. Closing of Trusted‑AI Panel

• Moderator thanked participants, highlighted the synergy of French tech + Indian scale, and invited the audience to mingle with speakers.

4. Transition & Brief Keynote on Multilingual AGI (Prof Raj Reddy)

• Prof Raj Reddy argued that India’s priority should be multilingual AGI to bridge the nation’s 22 official languages.
• Stressed the need for quantifiable metrics, edge‑centric sovereign models, and privacy‑preserving personal AI companions.
• Warned against focusing on generic AGI (e.g., “trillion‑dollar AGI”) that may be out‑sourced abroad.
• Highlighted AI for the bottom‑of‑the‑pyramid, citing the need for accessible, low‑cost AI tools for rural populations.

5. Panel – AI for Science

Moderator: Prof Abhay Karandikar (Secretary, DST, India)

5.1. Panel Composition

• Amit Shait – Founder, Indian AI Research Organisation (AIRO).
• Antoine Petit – CEO & Chairman, CNRS France.
• Joelle Pino – Chief AI Officer (organisation not specified).
• Irakli Baris – Head, Center of AI & Robotics, UNICRI (UN).
• Additional voices: Prof Zuhalpino (academic & industry AI leader), Prof R. Shait, and others.

5.2. Themes & Insights

Theme	Key Contributions
Building High‑End Talent & Indigenous Products (Amit Shait)	Emphasised the need for Indian‑born AI products, not just services. Described AIRO’s ecosystem: PhD‑training, industry‑university collaboration, IP & licensing pipelines, and seed‑to‑growth investment. Cited the lack of global Indian AI products as a gap to fill.
AI‑Enabled Scientific Discovery & Reproducibility (Antoine Petit)	Explained how AI reverses the material‑discovery workflow (design‑to‑properties). Stressed inter‑disciplinary cooperation, high‑performance computing, and the risk of AI‑generated false papers. Proposed a virtual “AI for Science” centre to foster collaboration.
AI as a Scientific Instrument (Joelle Pino)	Described AI as analogous to a new laboratory instrument that accelerates hypothesis generation, ranking of candidates, and verification cycles. Cited a crystallography project where AI reduced a 20‑year effort to one year by ranking candidate crystals. Highlighted transparency and evaluation standards as essential for reproducibility.
Responsible AI for Law‑Enforcement & Policy (Irakli Baris)	Presented UNICRI’s toolkit for responsible AI use in law‑enforcement, now piloted in India, Kazakhstan, Nigeria, Oman, Brazil. Discussed the dual‑use risk (criminal misuse vs. public‑trust) and the need for global policy frameworks that are “as smart as the technology”.
Ethics, Sovereignty & Global Collaboration (Various)	Panelists concurred that AI sovereignty ≠ isolation; open‑science, open‑software, and shared ethical charters are vital. Emphasised North‑South collaborations to bring AI to climate, agriculture, and health for countries lacking large experimental facilities.
Open‑Source vs. Closed‑Source Scientific Models (Discussion)	Prof Shait defended open‑source scientific models (citing the LAMA series) and warned that commercial fine‑tuning often stays closed, potentially stifling scientific progress.

5.3. Audience Q&A (selected)

• Question on Open‑Source Scientific Foundation Models – Prof Shait argued that open‑source accelerates reproducibility and that courageous industry players can defy the “keep‑it‑closed” trend.
• Question on Global Guidelines for Responsible AI – Irakli Baris referenced the UN‑endorsed responsible‑AI toolkit for law‑enforcement, now in operational use in India, as a concrete example of global governance.

5.4. Closing of AI‑for‑Science Panel

• Panelists reiterated the importance of pragmatic, use‑case‑driven research, transparent evaluation, and building a shared ethical foundation for AI‑driven scientific breakthroughs.

6. Final Wrap‑Up

• Moderator thanked speakers, announced a group photograph, and acknowledged platinum, gold, and silver sponsors (CMS, EGM Total, BNP Paribas, Capgemini, Schneider Electric, MBDA).
• Mark Villalmopillier delivered a brief momentum presentation, after which the session formally closed.

Key Takeaways

• Franco‑Indian AI partnerships are already materialising through dozens of signed agreements spanning AI, space, healthcare and industrial innovation.
• Trust is the prerequisite for AI scale; it must be baked into the architecture, data lineage, security, explainability, accountability and governance of every system.
• French deep‑tech (quantum, AI agents, photonics) complements India’s massive market and talent pool, enabling a “trust‑enabled” bilateral innovation engine.
• Leadership and culture matter: senior executives must model AI adoption, embed trust from the top, and foster an ecosystem mindset across industry, academia and government.
• AI for science is a new instrument that can compress discovery cycles (e.g., crystal‑ranking) but requires transparent data, reproducible benchmarks, and robust evaluation.
• Open‑source scientific models (e.g., LAMA, Merlin) enhance reproducibility and democratize access, while proprietary fine‑tuning may hinder collective progress.
• Responsible AI governance is advancing through UN‑backed toolkits and national pilots (India, Brazil, etc.), illustrating a path toward global standards and public trust.
• Multilingual AGI is identified as a uniquely Indian priority: building sovereign, edge‑centric AI companions that support all 22 official languages and bridge the digital divide.
• Mind‑set shift – from siloed, accuracy‑only AI to ecosystem‑wide, people‑in‑the‑loop, simulation‑backed, lifecycle‑governed AI – is essential for both trusted deployment and scientific impact.

---

See Also:

• ai-innovators-exchange-accelerating-innovation-through-startup-and-industry-synergy
• the-role-of-science-in-international-ai-governance
• democratizing-ai-resources-in-india
• democratizing-ai-resources-and-building-inclusive-ai-solutions-for-india
• building-resilient-sustainable-ai-infrastructure-for-people-planet-and-progress
• from-evidence-to-scale-testing-financing-and-operationalizing-technology-and-ai-for-development-and-humanitarian-action
• panel-on-the-2026-international-ai-safety-report