Sysdig introduces AI workload security to tackle active AI risks – iTWire

COMPANY NEWS: Sysdig, a leading provider of cloud security driven by real-time insights, has unveiled AI Workload Security, a new capability aimed at identifying and managing active risks associated with AI environments. This addition to Sysdig’s cloud-native application protection platform (CNAPP) is tailored to empower security teams to gain visibility into their AI workloads, spot suspicious activity in real time, and ensure compliance with emerging AI guidelines.

Sysdig senior vice president of product management Knox Anderson emphasised the necessity of this solution amidst the rising demand for secure AI adoption. Anderson highlighted how AI Workload Security enables organisations to comprehend their AI infrastructure, pinpoint active risks such as publicly exposed AI packages, and swiftly address vulnerabilities. Given that AI workloads are increasingly targeted by malicious actors, the new offering equips defenders with the tools to detect and mitigate threats to AI models and training data effectively.

The prevalence of Kubernetes as the preferred deployment platform for AI underscores the importance of securing containerised workloads. Sysdig’s CNAPP, built upon the open-source Falco, offers real-time visibility into cloud-native runtime security, regardless of deployment location—be it cloud or on-premises.

Real-time AI Workload Security from Sysdig enables companies to identify and prioritise workloads housing leading AI engines and software packages like OpenAI, Hugging Face, TensorFlow, and Anthropic. By understanding where AI workloads operate, Sysdig empowers organisations to manage and control AI usage, streamlining triage and response times through seamless integration with the unified risk findings feature.

Rising AI adoption heightens security concerns
Sysdig’s research reveals that 34% of deployed generative AI workloads are publicly exposed, posing imminent risks to sensitive data leveraged by these models. This widespread public exposure not only increases the likelihood of security breaches and data leaks but also invites regulatory compliance challenges.

The announcement of AI Workload Security comes at a pivotal moment as organisations increasingly embrace AI deployment while grappling with security concerns. According to a recent survey by the Cloud Security Alliance, 55% of organisations plan to implement generative AI solutions this year. Sysdig’s findings further indicate a significant uptick in the deployment of OpenAI packages since December.

Moreover, Sysdig’s AI Workload Security aligns with forthcoming guidelines and regulatory pressures on AI auditing and regulation, as outlined in the Biden Administration’s executive order from October 2023 and subsequent recommendations from the National Telecommunications and Information Administration (NTIA) in March 2024. By addressing public exposure, exploitable vulnerabilities, and runtime events, Sysdig’s solution aids organisations in swiftly addressing issues ahead of impending AI legislation.

Anderson emphasised the critical need for enhanced security controls and runtime detections tailored to the unique challenges posed by AI workloads. Sysdig’s AI Workload Security aims to help customers navigate these challenges effectively, safeguarding against potential threats while harnessing the efficiency and speed benefits offered by generative AI.

AI Workload Security will be available for all customers in May 2024, marking a significant step forward in fortifying AI environments against active risks.