Google has reinforced its stance that AI safety must function as core infrastructure if artificial intelligence is to achieve its full transformative potential. This message took centre stage at Google’s ‘Safe and Trusted AI’ event in New Delhi, held ahead of the upcoming AI Impact Summit 2026. The company announced that its AI safety roadmap will prioritise protections for children, teenagers and the elderly — groups often most vulnerable to online risks.
Safety as the Foundation of AI’s Future
Evan Kotsovinos, Google’s Vice President for Privacy, Safety and Security, said AI can only scale confidently into crucial sectors like healthcare, finance, and education if safety is treated as a fundamental building block.
In a country like India, where digital adoption is soaring, Kotsovinos noted that Google is building AI systems that actively protect user trust during the nation’s AI-driven transformation. He emphasised that AI-powered, on-device defences can help shield individuals and public infrastructure from cyber threats — a critical need in a country as large and diverse as India.
“India will lead the way in shaping equitable and reliable AI for the entire Global South,” he added.
Google India’s 360° Approach to Safety
Preeti Lobana, Google India’s Country Manager, outlined the company’s “360-degree” safety strategy, which includes:
- Built-in product and Cloud protections
- Digital literacy initiatives
- Ongoing research investments
- Collaboration with partners across the tech ecosystem
She stressed that Google’s objective is simple: make safety the default across the entire digital environment.
Stronger Protections for Kids, Teens, and the Elderly
A major portion of Google’s safety work focuses on its most vulnerable users. New on-device features — faster, private, and always-on — are designed to protect against scams and misinformation.
Real-time Scam Call Alerts (Powered by Gemini Nano)
Rolling out on Pixel devices, this feature:
- Analyses unknown calls in real time
- Flags suspicious behaviour
- Works fully on-device without recording audio
- Can be turned off at any time
Screen-Share Scam Protection (Pilot)
A new safeguard for UPI and banking app users alerts them instantly if they:
- Are screen sharing and
- Have financial apps like Google Pay, Navi, or Paytm open
The user can stop sharing or disconnect the call with one tap — an essential defence against live fraud attempts.
SynthID: Expanding AI Content Authenticity Tools
Google is expanding access to SynthID, its AI-generated content detection technology.
New updates include:
- SynthID Detector availability for early partners
- Open-sourcing of SynthID text watermarking via the Responsible GenAI Toolkit
Media groups including Jagran, PTI, and India Today are among the early adopters.
Boosting Digital Literacy Across India
Google will launch its Learn and Explore Online (LEO) programme in India in December 2025. LEO will train teachers, parents, and practitioners on:
- Parental controls
- Age-appropriate online experiences
Additional initiatives include:
- A $200,000 Google.org grant to the CyberPeace Foundation to support AI-driven cyber defence and safer digital learning
- “Sach Ke Sathi, DigiKavach for Seniors”, offering in-person training to 5,000+ seniors across 30 cities
Google is also working with IIT Madras and CeRAI to advance AI safety research tailored to India’s languages and device diversity.
Building the Future of Safe AI
Google showcased several advanced tools that secure users and developers:
- CodeMender: Automatically identifies and patches zero-day vulnerabilities
- Private AI Compute, Parfait, VaultGemma: Privacy Enhancing Technologies for safer AI models
- Enhanced Phone Number Verification (ePNV): A new SIM-based alternative to vulnerable SMS OTPs