Data Destruction via Indirect Prompt Injection Targeting Claude Computer-Use - AI Case Study

When deploying tools that will be shared across multiple AI agents, it is important to implement robust policies and controls on permissions for the tools. These controls include applying the principle of least privilege along with delegated access, where the tools receive the permissions, identities, and restrictions of the AI agent calling them. These configurations may be implemented either in MCP servers which connect the agents to the tools calling them or, in more complex cases, directly in the configuration files of the tool.

Implement logging of inputs and outputs of deployed AI models. When deploying AI agents, implement logging of the intermediate steps of agentic actions and decisions, data access and tool use, installation commands, and identity of the agent. Monitoring logs can help to detect security threats and mitigate impacts. Additionally, having logging enabled can discourage adversaries who want to remain undetected from utilizing AI resources.

Guardrails are safety controls that are placed between a generative AI model and the output shared with the user to prevent undesired inputs and outputs. Guardrails can take the form of validators such as filters, rule-based logic, or regular expressions, as well as AI-based approaches, such as classifiers and utilizing LLMs, or named entity recognition (NER) to evaluate the safety of the prompt or response. Domain specific methods can be employed to reduce risks in a variety of areas such as etiquette, brand damage, jailbreaking, false information, code exploits, SQL injections, and data leakage.

Guidelines are safety controls that are placed between user-provided input and a generative AI model to help direct the model to produce desired outputs and prevent undesired outputs. Guidelines can be implemented as instructions appended to all user prompts or as part of the instructions in the system prompt. They can define the goal(s), role, and voice of the system, as well as outline safety and security parameters.

When training or fine-tuning a generative AI model it is important to utilize techniques that improve model alignment with safety, security, and content policies. The fine-tuning process can potentially remove built-in safety mechanisms in a generative AI model, but utilizing techniques such as Supervised Fine-Tuning, Reinforcement Learning from Human Feedback or AI Feedback, and Targeted Safety Context Distillation can improve the safety and alignment of the model.

Systems should require the user or another human stakeholder to approve AI agent actions before the agent takes them. The human approver may be technical staff or business unit SMEs depending on the use case. Separate tools, such as dedicated audit agents, may assist human approval, but final adjudication should be conducted by a human decision-maker. The security benefits from Human In-the-Loop policies may be at odds with operational overhead costs of additional approvals. To ease this, Human In-the-Loop policies should follow the degree of consequence of the task at hand. Minor, repetitive tasks performed by agents accessing basic tools may only require minimal human oversight, while agents employed in systems with significant consequences may necessitate approval from multiple stakeholders diversified across multiple organizations.

Implement validation on inputs and outputs for the tools and data sources used by AI agents. Validation includes enforcing a common data format, schema validation, checks for sensitive or prohibited information leakage, and data sanitization to remove potential injections or unsafe code. Input and output validation can help prevent compromises from spreading in AI-enabled systems and can help secure the workflow when multiple components are chained together. Validation should be performed external to the AI agent.

AI agents may be granted elevated privileges above that of a normal user to enable desired workflows. When deploying a privileged AI agent, or an agent that interacts with multiple users, it is important to implement robust policies and controls on permissions of the privileged agent. These controls include Role-Based Access Controls (RBAC), Attribute-Based Access Controls (ABAC), and the principle of least privilege so that the agent is only granted the necessary permissions to access tools and resources required to accomplish its designated task(s).

Untrusted data can contain prompt injections that invoke an AI agent's tools, potentially causing confidentiality, integrity or availability violations. It is recommended that tool invocation be restricted or limited when untrusted data enters the LLM's context. The degree to which tool invocation is restricted may depend on the potential consequences of the action. Consider blocking the automatic invocation of tools or requiring user confirmation once untrusted data enters the LLM's context. For high consequence actions, consider always requiring user confirmation.

Define security boundaries around agentic tools and data sources with methods such as API access, container isolation, code execution sandboxing, and rate limiting of tool invocation. When sandboxing, limit resource and network access and build the container or virtual machine from a clean base image before each run. This restricts untrusted processes or potential compromises from spreading throughout the system.

When deploying an AI agent that acts as a representative of a user and performs actions on their behalf, it is important to implement robust policies and controls on permissions and lifecycle management of the agent. Lifecycle management involves establishing identity, protocols for access management, and decommissioning of the agent when its role is no longer needed. Controls should also include the principle of least privilege and delegated access from the user account. When acting as a representative of a user, the AI agent should not be granted permissions that the user would not be granted within the system or organization.