TASI 2026: AI Safety and Child Protection Take Centre Stage
Industry leaders and policymakers debate the urgent need for AI governance and safety.
1AI Safety Lags Behind Innovation
Technology and policy experts gathered at the TASI Festival in New Delhi on October 14 to address the widening gap between AI capabilities and safety protocols. Executives from OpenAI, Google, and Roblox debated whether voluntary industry collaboration is sufficient or if governments must mandate stricter regulatory frameworks before public deployment.
2Incentives Versus Market Competition
While companies maintain an inherent incentive to foster user trust, intense market competition often diverts resources away from safety initiatives. Rohit Kumar of The Quantum Hub argued that because the AI landscape moves rapidly, regulation is necessary to align corporate incentives with public safety, ensuring long-term stability for the ecosystem.
3Rigorous Red-Teaming and Testing
Google’s Snigdha Bhardwaj emphasized a zero-tolerance policy regarding harmful content, particularly CSAM. The company utilizes continuous red-teaming to identify vulnerabilities before any model release. Executives proposed an interoperable safety layer across platforms to share risk signals, which would reduce competitive barriers while effectively addressing privacy and intellectual property concerns.
4Prioritizing Child Safety Defaults
Nicky Jackson Colaco of Roblox argued that child safety must be a default setting rather than a burden placed on parents. The panel discussed implementing robust age-assurance measures and protective experiences that automatically trigger when a user's age is uncertain, pushing for industry-wide standards that prioritize the protection of younger digital users.
5Impact on Cognitive Development
Akash Pugalia highlighted concerns regarding the erosion of critical thinking skills in children who rely heavily on AI for answers. Furthermore, the increasing sophistication of audio and video deepfakes poses significant risks, making it difficult for users to distinguish between authentic content and manipulated media, which complicates digital literacy efforts.
6Adapting AI for Local Contexts
OpenAI’s Pragya Misra noted that safety systems must account for India’s unique multi-cultural landscape and linguistic diversity. To prevent AI from undermining the learning process, the company introduced a Study Mode that utilizes Socratic methodology, guiding students toward answers rather than providing them directly, thus maintaining academic rigor during digital interactions.
7Regulatory Gaps in India
Rohit Kumar criticized India’s current reliance on safe harbor laws, which primarily focus on third-party content removal rather than modern AI governance. He advocated for non-prescriptive, flexible policies similar to the UK’s online safety laws, which establish clear norms for child protection without stifling innovation or restricting freedom of expression.
8Future of AI Governance
The TASI Festival highlighted that effective AI governance requires a collaborative approach between industry, policymakers, and civil society. As AI continues to evolve, the consensus remains that safety must be integrated into the development lifecycle, ensuring that technological progress benefits humanity while mitigating risks associated with deepfakes, privacy, and child safety.