https://www.infoq.com InfoQ QCon San Francisco 2025 Articles feed https://www.infoq.com/articles/secure-ai-development/?utm_campaign=infoq_content&utm_source=infoq&utm_medium=feed&utm_term=QCon+San+Francisco+2025-articles <img src="https://res.infoq.com/articles/secure-ai-development/en/headerimage/header-1765541260371.jpg"/><p>Autonomous AI agents amplify productivity but can cause severe damage without safeguards. Defend the ReAct loop—context, reasoning, and tools—through provenance gates, planner-critic separation, scoped credentials, sandboxed code, and STRIDE/MAESTRO threat modeling. With robust logging, bounded autonomy, and red-teaming, agents can deliver trustworthy productivity while minimizing risk.</p> <i>By Sriram Madapusi Vasudevan</i> Application Security QCon Software Development Conference Agents QCon San Francisco 2025 Architecture Security AI Architecture Architecture & Design article Tue, 16 Dec 2025 09:00:00 GMT https://www.infoq.com/articles/secure-ai-development/?utm_campaign=infoq_content&utm_source=infoq&utm_medium=feed&utm_term=QCon+San+Francisco+2025-articles Sriram Madapusi Vasudevan 2025-12-16T09:00:00Z /articles/secure-ai-development/en