AWS for Industries
Aligning AWS to MPA security best practices for media archives – Part 3
This is part 3 of our three-part series on how Amazon Web Services (AWS) aligns with MPA Content Security Best Practices with a particular focus on media archives. In part 1, we covered foundational controls like information security management systems (ISMS), authentication, authorization, network isolation, and encryption. In part 2, we addressed operational controls like patch management, monitoring, personnel security, data integrity, and business continuity. In this final installment, we explore the newest additions to the MPA Best Practices, which cover Zero Trust Architecture (ZTA), artificial intelligence and machine learning (AI/ML) security governance, cloud misconfiguration prevention, secure content transfer, and change control.
These controls reflect the industry’s evolving threat landscape. Studios and broadcasters now face AI-powered content scraping, increasingly sophisticated supply chain attacks, and the challenge of securing distributed cloud workloads that span multiple providers and geographic boundaries. MPA v5.3 responds with five new topics and controls that AWS is uniquely positioned to address.
Zero Trust Architecture
MPA v5.3 introduced ZTA as a dedicated control family that is a significant addition reflecting the industry’s shift away from perimeter-based security. Control TS-6.0 requires authenticating each user and system before workflow access. Control TS-6.1 mandates that all connections are deny-by-default authorized by a policy-based centrally stored authorization engine using attributes such as identity, device, location, and data sensitivity. Control TS-6.2 requires organizations to define, map, and regularly review all protect surfaces like the critical data, applications, assets, and services that demand the highest security.
AWS Verified Access delivers network-level zero trust by validating every access request against identity and device trust policies before granting connectivity. This eliminates the need for VPNs for application access. Amazon VPC Lattice provides service-to-service connectivity with built-in authentication and authorization policies, enforcing TS-6.1’s requirement for policy-based access decisions at every connection.
AWS IAM Identity Center centralizes workforce identity with contextual access controls. For protect surface mapping (TS-6.2), AWS Resource Explorer discovers and catalogs all resources across accounts, and AWS Security Hub resource findings identify which assets carry security risks. Virtual private cloud (VPC) security groups enable micro-segmentation around each protect surface, restricting lateral movement to only explicitly permitted traffic flows.
AI/ML security management
Also new to v5.3, controls OR-5.0, TS-7.0, and TS-7.1 establish formal AI/ML security governance. Those directly respond to studio concerns about content being inadvertently used for model training. OR-5.0 requires a formal AI/ML security management policy covering content protection, acceptable use, data handling, and security reviews. TS-7.0 mandates scanning of all pre-trained models and datasets for malicious content and adherence to data I/O workflow principles. TS-7.1 requires secure configuration of AI/ML applications including access controls, logging, and content boundary enforcement.
AWS provides multiple layers of AI/ML governance. By using AWS AI service opt-out policies in AWS Organizations, you can explicitly prevent your content from being used to improve AWS AI services to directly respond to studio concerns about content leakage.
Amazon Bedrock Guardrails enforces content filters, topic boundaries, and sensitive information protection on AI model inputs and outputs. For TS-7.0’s scanning requirements, Amazon SageMaker Model Cards document model provenance and risk ratings, and ML Governance with Amazon SageMaker provides model access controls, approval workflows, and audit trails. For isolated processing, a key architectural control under TS-7.0, SageMaker supports network isolation mode, where training and inference containers run without internet access, preventing data exfiltration from AI/ML workloads processing prerelease content.
For TS-7.1’s requirement to secure AI/ML application configurations, Amazon Bedrock AgentCore provides a defense-in-depth architecture for AI agents. AgentCore Identity authenticates agent-to-tool interactions, so only authorized agents access sensitive media workflows. AgentCore Policy uses Cedar-based declarative policies on AgentCore Gateway to enforce fine-grained authorization, controlling which tools an agent can invoke, under what conditions, and with what data boundaries. For configuration governance, AWS Config rules validate AI/ML resources against defined baselines, and AWS Systems Manager enforces platform-level settings.
This creates a protective boundary around agent operations that maps directly to TS-7.1’s content boundary enforcement requirements.
Finally, AWS CloudTrail data events for Amazon Bedrock capture every model invocation for compliance logging.
Cloud misconfigurations and shared responsibility
Controls TS-1.11 and TS-1.12 address the most common cause of cloud security incidents: misconfigurations. TS-1.11 requires formal governance between cloud service providers (CSPs) and cloud service consumers (CSCs), including regular compliance validation and contractual security obligations. TS-1.12 mandates proactive detection and correction of cloud misconfigurations with alerting, defined roles for remediation, and configuration management tooling.
AWS offers a defense-in-depth approach to preventing misconfigurations. AWS Config provides over 300 managed rules that constantly evaluate resource configurations against best practices, detecting Amazon Simple Storage Service (Amazon S3) buckets without encryption, security groups with overly permissive rules, or unencrypted Amazon Elastic Block Store (Amazon EBS) volumes. AWS Security Hub CSPM runs automated checks against Center for Internet Security (CIS) AWS Foundations Benchmark and AWS Foundational Security Best Practices, scoring your environment’s security posture. Amazon S3 Block Public Access, which is enabled by default for all new buckets since April 2023, prevents accidental public exposure of archive content.
AWS Identity and Access Management Access Analyzer identifies resources shared with external entities, directly supporting TS-1.11’s governance requirements. For the shared responsibility model itself, the AWS Customer Compliance Guides are available on the Compliance Resources page and document exactly which controls AWS manages versus customer responsibilities.
Secure content transfer and data I/O
Control TS-1.0 requires dedicated, isolated systems for moving content between external and internal networks, malware scanning, and approval gates, with documented procedures. Control TS-1.16 mandates dedicated content transfer systems with approval processes, separate transfer and production environments, transfer logging, and end-to-end encryption.
The AWS Transfer Family provides fully managed SFTP, FTPS, FTP, and AS2 endpoints, protocols commonly used in studio supply chains, with integration into Amazon S3 and Amazon Elastic File System (Amazon EFS) for storage.
Amazon S3 presigned URLs with time-limited access tokens enable secure content delivery without exposing bucket credentials. AWS DataSync accelerates secure bulk transfers with built-in integrity validation, encryption in transit, and detailed transfer logging, satisfying TS-1.16’s audit requirements.
For malware scanning, Amazon GuardDuty Malware Protection for S3 automatically scans newly uploaded objects and quarantines threats before they reach production workflows, addressing TS-1.0’s requirement for scanning content during transfer. Amazon Macie identifies sensitive data patterns in S3 objects, and GuardDuty S3 Protection detects anomalous data access patterns that could indicate unauthorized exfiltration. Approval gates, as required for TS-1.0, can be modeled using AWS Step Functions based approval workflows.
Change control and secure SDLC
Control TS-5.0 requires formal change control with an up-to-date inventory of components, approval workflows for all changes, and testing in non-production environments before deployment. Controls TS-1.13 and TS-1.14 mandate secure software development lifecycle (SSDLC) practices including code reviews, security testing, and Secure by Design and Privacy by Design principles across the development lifecycle.
AWS CloudFormation and AWS Cloud Development Kit (AWS CDK) enable infrastructure as code (IaC) with version-controlled, peer-reviewed changes, satisfying TS-5.0’s approval and documentation requirements. AWS Service Catalog provides preapproved, security-hardened resource configurations that enforce the Secure by Design principle. Amazon Inspector scans Amazon Elastic Compute Cloud (Amazon EC2) instances, container images, and AWS Lambda functions for known Common Vulnerabilities and Exposures (CVEs) in deployed workloads. For deeper validation, AWS Security Agent goes beyond static analysis by testing running applications with multistep attack scenarios that account for business context to comply with TS-1.13’s requirement that code is reviewed and tested before each release.
Conclusion
With part 3 covering the newest MPA v5.3 control families, this series demonstrates broad alignment between AWS services and the full breadth of MPA requirements for media archives. Zero Trust Architecture (TS-6.0 through TS-6.2) maps to AWS Verified Access, VPC Lattice, and IAM Identity Center. AI/ML security governance (OR-5.0, TS-7.0, TS-7.1) is addressed through AI service opt-out policies, Amazon Bedrock Guardrails, and ML Governance in SageMaker. Cloud misconfiguration prevention (TS-1.11, TS-1.12) is automated through AWS Config, Security Hub, and IAM Access Analyzer. Secure content transfer (TS-1.0, TS-1.16) maps to Transfer Family, DataSync, and presigned URLs. And change control (TS-5.0, TS-1.13, TS-1.14) is enforced through CloudFormation, AWS CodePipeline, and AWS CodeGuru.
Together with the foundational controls in part 1 and the operational controls in part 2, media organizations can build archive environments that satisfy the most demanding studio security requirements while benefiting from the operational efficiency of a modern cloud architecture.
To get started, explore the AWS MPA compliance page and review the MovieLabs CSAP alignment series for prescriptive implementation guidance.