AWS releases TOLAP for object-level access control in AI agents
Key Takeaway
TOLAP provides a standardized way to enforce fine-grained data security, such as row and column filtering, directly at the tool level for AI agents.
Summary
AWS has open-sourced the Tool-Object Level Access Protocol (TOLAP) under the Apache-2.0 license. The project addresses a security gap where AI agents can access unauthorized data through tools even when IAM or OAuth are active. TOLAP enables fine-grained control over specific data objects like rows, columns, and fields across databases, APIs, and object storage.
Technical components of TOLAP
- Versioned policy schema covering databases, APIs, knowledge bases, and object storage.
- Enforcement SDKs for .NET, Python, and TypeScript with zero external dependencies in core packages.
- Reference policy server and an authoring console for policy management.
- Fourteen integrations including LangChain, Semantic Kernel, Bedrock Agents, and OpenAI Agents.
- Most-restrictive-wins merging logic for handling multiple overlapping policies.
- Source-point enforcement that applies security at the data origin rather than the model output layer.
Why it matters
TOLAP provides a standardized security layer for developers building AI agents, preventing prompt injection from exposing sensitive data that has already entered the model's context window.
Original reference: aws.amazon.com →