Shared Responsibility and the Foundations of AI Security

AWS Certified AI Practitioner · Security, Compliance, and Governance for AI Solutions · By Super Admin

The AWS shared responsibility model divides security duties between AWS and customers, and AI security addresses threats to AI confidentiality, integrity, and availability across three categories of AI application.

AWS prioritizes security above all else, providing infrastructure and services that meet even the strictest security standards, but keeping an AWS environment secure requires AWS and the customer to work together. This is the shared responsibility model, and it applies to every AWS service, including AI services.

The AWS Shared Responsibility Model

Security works in two key ways. AWS handles security "of the cloud" by maintaining and protecting the core infrastructure, such as data centers and networks, that powers its services; independent auditors regularly verify these measures through compliance programs. The customer manages security "in the cloud," meaning they are responsible for how they use AWS services securely. The extent of a customer's security duties depends on which services they choose, their data's sensitivity, their organization's policies, and any relevant regulations.

Defining AI Security

AI technology is growing quickly and being used more widely; once found only in controlled environments for specific purposes, AI-enabled systems are now integrated into many applications accessible to the general public. AI security refers to the techniques, approaches, and procedures that detect and block threats targeting an AI model or AI-enabled system's confidentiality, integrity, or availability. It is essential throughout AI development to ensure systems function safely and reliably during operation. Beyond conventional cybersecurity vulnerabilities, integrating AI creates additional attack vectors and weaknesses requiring specialized security measures, and identifying and addressing these AI-specific vulnerabilities is fundamental to AI security, demanding both technical and operational solutions.

Three Categories of AI in Applications

The global conversation frequently groups various algorithmic technologies together as "AI," which makes clear analysis difficult. The AI ecosystem can be better understood by dividing it into three categories based on how AI appears in applications. The first is engineered systems incorporating AI components or subsystems, such as smart home devices optimizing energy use, email spam filters, streaming recommendation algorithms, and medical diagnostic imaging tools. The second is AI that enhances human capabilities, such as AI coding assistants, AI-powered design tools, language translation apps, and surgical robots guided by surgeons. The third is AI operating independently with its own agency, such as autonomous vehicles, AI trading systems executing transactions without approval, smart manufacturing robots, and AI security systems that respond to threats without human intervention. Making these distinctions is crucial because each category faces different threats and risks and requires different approaches to threat mitigation and risk management.

Key takeaways

  • AWS secures 'of the cloud' infrastructure while customers secure 'in the cloud' usage, and this applies to AI services too.
  • AI security detects and blocks threats to AI model/system confidentiality, integrity, and availability, requiring specialized measures beyond conventional cybersecurity.
  • AI appears in applications as engineered AI components, AI enhancing human capability, or AI with independent agency, each facing different risks.
0.0 (0 ratings)

Discussion (0)

No discussion on this article yet — be the first to comment.

Please sign in to join the discussion.