Google puts kids, teens and elderly at the centre of Global AI safety

Google has asserted that safety must be treated as core infrastructure for Artificial Intelligence (AI) to deliver its full, transformative impact. This was the central focus of the ‘Safe and Trusted AI’ event hosted by Google in the capital, held ahead of the upcoming AI Impact Summit 2026. The company is explicitly placing the protection of kids, teens, and the elderly at the heart of its AI Safety efforts.

Evan Kotsovinos, Vice President, Privacy, Safety and Security at Google, stated that safety is the fundamental infrastructure allowing AI to move confidently into areas like finance, healthcare, and education, thereby forming the basis for economic growth. He highlighted that in India, where the digital economy is currently booming, Google is building AI systems designed to maintain user trust during the country's transition to AI. Mr Kotsovinos noted that AI-powered defences running on the device offer an unprecedented advantage against bad actors who seek to target individuals or critical public infrastructure. He emphasised that India’s scale and multiplicity make it the world’s proving ground for reliable AI safety, concluding that the path to equitable AI for the entire Global South will be led from India.

Preeti Lobana, Country Manager, Google India, detailed the company’ "360-degree" approach to safety. She explained this involved combining on-product and on-Cloud protections, promoting digital literacy, investing in continuous safety research, and collaborating with the wider ecosystem. The core idea, she summarised, is to build an ecosystem where safety is embedded by default.

Protecting children, teens and the elderly across its platforms is a strong priority at Google in the area of safety. The company is bringing protections and features that run directly on the device, making them faster, more private and always on, along with investing in digital literacy efforts to close the gap from both ends.

This feature is powered by Gemini Nano and is rolling out on Pixel phones. It analyses calls in real-time and flags potential scams entirely on the device, without recording audio, transcripts, or sharing data with Google. The feature is off by default, applies only to calls from unknown numbers, and can be turned off by the user at any time.

This new pilot feature will show crucial alerts when users are at high risk. If a user screen-shares while speaking to an unknown caller and opens financial apps such as Google Pay, Navi, or Paytm, the phone will display an immediate warning. The user can then stop the screen-sharing and the call with a single tap. The feature is specifically designed to prevent scams that target UPI and banking apps during live calls.

Google is broadening the availability of its SynthID Detector and releasing the SynthID text watermarking tool as an open-source component of its Responsible GenAI Toolkit. This watermarking technology helps academic, research, and publishing partners identify AI-generated images and audio. Strategic partners, including academia, researchers, and publishers such as Jagran, PTI, India Today and more are receiving early access to the SynthID Detector and API.

Google is also investing heavily in digital literacy efforts and strengthening the broader technology ecosystem, bringing its flagship Learn and Explore Online (LEO) programme to India in December 2025, which trains teachers, practitioners, and parents on using parental tools and creating age-appropriate online experiences. The company is providing a $200,000 USD grant from Google.Org to the CyberPeace Foundation in India. This funding will support capacity building, deliver AI-driven cyber-defence tools, and create safer digital learning environments for children and teens. Additionally, the "Sach Ke Sathi, DigiKavach for Seniors” programme offers in-person training to over 5,000 seniors in roughly 30 cities.

In partnership with the Reserve Bank of India (RBI), Google has helped publish a verified list of authorised digital lending apps and their associated NBFCs to aid users in identifying trustworthy applications. The company is also collaborating with IIT Madras and CeRAI (Centre for Responsible AI) to advance AI safety research that reflects India’s diverse language and device ecosystem.

Google highlighted CodeMender, a self-defending security system that autonomously identifies and patches zero-day vulnerabilities in open-source software. Recent releases like Private AI Compute, Parfait, and VaultGemma are also Privacy Enhancing Technologies (PETs) designed to help the wider ecosystem build privacy-preserving AI systems. The company is also developing Enhanced Phone Number Verification (ePNV), a new technology aiming to replace less secure SMS-based one-time passwords with a SIM-based verification check to strengthen sign-in security.