Cisco has introduced AI Defense, a system designed to protect AI systems from attacks and misuse, ensuring the secure behavior of AI models. This software is specifically developed for businesses to reliably create, deploy, and operate AI applications. According to the manufacturer, AI Defense integrates into the network structure to detect and fend off threats during the development and access of AI applications. DJ Sampath, Cisco’s Vice President of Product, AI Software, and Platform, stated that it is a reality that “we live in a multi-model and multi-cloud world.” Therefore, AI Defense is applicable regardless of the AI model used and the location of data processing.
Cisco views the increasing diversity in models and locations as a major security challenge, especially when training occurs with “non-public data” on third-party models. AI Defense aims to establish guidelines for the development and application of AI. According to Cisco’s AI Readiness Index, only 29% of respondents feel fully equipped to detect and prevent unauthorized manipulations with AI. AI Defense is designed to protect against attacks on AI and secure model behavior across platforms. It will block applications used as shadow AI in public and private clouds based on a set of rules and prevent data leaks by securing access to specific AI applications.
AI Defense uses automated tests to validate AI models against hundreds of potential security risks. This AI-driven review of algorithms identifies vulnerabilities and provides recommendations to security teams. Continuous validation during the runtime of AI applications aims to prevent prompt injections, denial-of-service attacks, and access to confidential data.
AI Defense implements an overarching security layer alongside existing model restrictions, ensuring consistent AI security policies regardless of the model used. The software employs machine learning based on data from Splunk (acquired by Cisco in 2023) and Cisco Talos to adapt to changes. The security mechanisms are also effective for apps containing only AI components.
Cisco’s significant investments in AI are bearing fruit with AI Defense, especially as the network business faced pressure recently. The increasing use of AI brings additional threat scenarios, and many are hesitant to use and develop AI products due to privacy and information security concerns. AI Defense addresses these issues by providing enhanced protection during the development and use of AI applications. The software aims to protect against the misuse of AI tools and data leaks within them. AI Defense is based on Cisco’s acquisitions of Armorblox and Robust Intelligence and is expected to be available from March.