
When an Authorised AI Agent Gets It Wrong — Aditi Sawhney, Mastercard
An AI agent can have the right credentials, the right permissions and authorisation for every step, and still do something nobody wanted. As agents take on more autonomy, the security question shifts from "who has access?" to "is this what the person actually meant?"
Aditi Sawhney, Senior Vice President of Security Solutions at Mastercard in Asia Pacific, explains what that means for payments and the rest of the enterprise.
What we cover:
- Agent Pay and Mastercard's trust framework for agentic payments
- Telling a legitimate agent from a rogue one, and why velocity checks fail
- Whether today's identity and authorisation standards are enough for agents
- Verifiable intent and interoperability across the industry • Who carries the liability when an authorised agent gets it wrong
- A CISO's three priorities in Asia Pacific for the next 12 months
Figures mentioned:
Mastercard, A Short History of the Future of Shopping and Payments (2026, forecasts 300 million shoppers routinely using AI agents by 2030);
McKinsey, The agentic commerce opportunity (2025, US B2C retail);
Verizon Data Breach Investigations Report 2025 and 2026.
Connect with Aditi Sawhney: https://www.linkedin.com/in/aditi-sawhney/
The Loop Asia: www.theloop.asia
Otros episodios de "The Loop Asia"



No te pierdas ningún episodio de “The Loop Asia”. Síguelo en la aplicación gratuita de GetPodcast.








