OpenAI Guardrails
Decide whether to block, log, or proceed with requests based on OpenAI Guardrail verdicts. Configure verdict actions like async checks, request denial, and feedback logging—tailored to your enforcement needs.
OpenAI Guardrails
OpenAI powers some of the most advanced language models in the world, including GPT-4, o3, Whisper, and DALL·E. These models are widely used for building applications across chat, summarization, code generation, document understanding, transcription, and multimodal tasks. But running them in production requires more than just model performance—it demands safety, compliance, and control.
Portkey acts as a powerful gateway layer for OpenAI, allowing you to apply customizable guardrails to every request without changing your application code. Whether you're concerned about prompt injection, data redaction, output filtering, or governance, Portkey provides a seamless way to manage and secure AI usage at scale.
With Portkey, you can:
Monitor and trace usage at a granular level
Apply safety filters and compliance checks to every prompt and response
Route requests intelligently based on cost, latency, or fallback rules
Set up rate limits, usage quotas, and custom rules per user, team, or feature
Gain real-time insights into model usage and performance
Portkey supports all OpenAI models out of the box and can be deployed in hours—not weeks—making it the easiest way to bring enterprise-grade control to your AI stack.
World-Class Guardrail Partners
Integrate top guardrail platforms with Portkey to run your custom policies seamlessly — from content filtering and PII detection to moderation and compliance. Ensure every AI request is safe, auditable, and aligned with your enterprise standards.
Mistral
Propmt Security
Patronus
Pillar
Lasso
Pangea
Bedrock
Azure
Lasso
Pangea
Bedrock
Azure
Promptfoo
Aporia
Acuvity
Exa
Guardrail checks you can apply to OpenAI
Guardrail checks you can apply to OpenAI
Guardrail checks you can apply to OpenAI
Portkey offers both deterministic and LLM-powered guardrails that work seamlessly with OpenAI’s APIs. You can apply these checks to inputs, outputs, or both.
Portkey offers both deterministic and LLM-powered guardrails that work seamlessly with OpenAI’s APIs. You can apply these checks to inputs, outputs, or both.
Portkey offers both deterministic and LLM-powered guardrails that work seamlessly with OpenAI’s APIs. You can apply these checks to inputs, outputs, or both.
Input guardrails
*
.
Regex Match
Basic
Enforce patterns on input prompts
Sentence / Word / Character Count
Basic
Control verbosity
Lowercase Detection
Basic
Control verbosity
Ends With
Basic
Validate specific prompt endings
Webhook
Basic
Enforce custom business logic
JWT Token Validator
Basic
Verify token authenticity
Model Whitelist
Basic
Allow only approved models per route
*
.
Moderate Content
★ Pro
Block unsafe or harmful prompts
Check Language
★ Pro
Enforce language constraints
Detect PII
★ Pro
Prevent sensitive info in prompts
Detect Gibberish
★ Pro
Block incoherent or low-quality input
Output guardrails
*
.
Regex / Sentence / Word / Character Count
Basic
Need content here Need content here
JSON Schema / JSON Keys
Basic
Ensure required words or phrases
Contains
Basic
Ensure required words or phrases
Valid URLs
Basic
Validate links in responses
Contains Code
Basic
Detect code in specific formats
Lowercase Detection / Ends With
Basic
Need content here Need content here
Webhook
Basic
Post-process or validate output
Detect PII / Detect Gibberish
Basic
Need content here Need content here
Partner Guardrails
Acuvity
Partner
Acuvity is a model-agnostic GenAI security solution. It is built to secure existing and future GenAI models, apps, services, tools, plugins, and more.Scan Content: Comprehensive content safety and security checks.
✓ Scan Content: Comprehensive content safety and security checks.
Acuvity
Partner
Acuvity is a model-agnostic GenAI security solution. It is built to secure existing and future GenAI models, apps, services, tools, plugins, and more.Scan Content: Comprehensive content safety and security checks.
✓ Scan Content: Comprehensive content safety and security checks.
Aporia
Partner
Aporia provides state-of-the-art Guardrails for any AI workload. With Aporia, you can set up powerful, multimodal AI Guardrails.
✓ Validate Project: Check all the policies within an Aporia project
Aporia
Partner
Aporia provides state-of-the-art Guardrails for any AI workload. With Aporia, you can set up powerful, multimodal AI Guardrails.
✓ Validate Project: Check all the policies within an Aporia project
Lasso Security
Partner
Lasso Security protects your GenAI apps from data leaks, prompt injections, and other potential risks, keeping your systems safe and secure.
✓ Analyse Content - Lasso Security's Deputies analyze content for various security risks, including jailbreak attempts, custom policy violations, sexual content, hate speech, illegal content, and more.
Lasso Security
Partner
Lasso Security protects your GenAI apps from data leaks, prompt injections, and other potential risks, keeping your systems safe and secure.
✓ Analyse Content - Lasso Security's Deputies analyze content for various security risks, including jailbreak attempts, custom policy violations, sexual content, hate speech, illegal content, and more.
Lasso Security
Partner
Lasso Security protects your GenAI apps from data leaks, prompt injections, and other potential risks, keeping your systems safe and secure.
✓ Analyse Content - Lasso Security's Deputies analyze content for various security risks, including jailbreak attempts, custom policy violations, sexual content, hate speech, illegal content, and more.
Mistral
Partner
Mistral moderation service helps detect and filter harmful content across multiple policy dimensions to secure your AI applications.
✓ Moderate Content: Checks if content passes selected content moderation checks
Mistral
Partner
Mistral moderation service helps detect and filter harmful content across multiple policy dimensions to secure your AI applications.
✓ Moderate Content: Checks if content passes selected content moderation checks
Pangea
Partner
Pangea AI Guard helps analyze and redact text to prevent model manipulation and malicious content.
✓ AI Guard - Analyze and redact text to avoid manipulation of the model and malicious content.
Pangea
Partner
Pangea AI Guard helps analyze and redact text to prevent model manipulation and malicious content.
✓ AI Guard - Analyze and redact text to avoid manipulation of the model and malicious content.
Azure Guardrails
Partner
Microsoft Azure offers robust content moderation and PII redaction services that can now be seamlessly integrated with Portkey’s Guardrails ecosystem.
✓ Azure Content Safety: A comprehensive content moderation service that detects harmful content, including hate speech, violence, sexual content, and self-harm references in text.
✓ Azure PII Detection: Advanced detection of personally identifiable information (PII) and protected health information (PHI) to safeguard sensitive data.
Azure Guardrails
Partner
Microsoft Azure offers robust content moderation and PII redaction services that can now be seamlessly integrated with Portkey’s Guardrails ecosystem.
✓ Azure Content Safety: A comprehensive content moderation service that detects harmful content, including hate speech, violence, sexual content, and self-harm references in text.
✓ Azure PII Detection: Advanced detection of personally identifiable information (PII) and protected health information (PHI) to safeguard sensitive data.
Azure Guardrails
Partner
Microsoft Azure offers robust content moderation and PII redaction services that can now be seamlessly integrated with Portkey’s Guardrails ecosystem.
✓ Azure Content Safety: A comprehensive content moderation service that detects harmful content, including hate speech, violence, sexual content, and self-harm references in text.
AWS Bedrock Guardrail
Partner
AWS Bedrock provides a comprehensive solution for securing your LLM applications, including content filtering, PII detection, redaction, and more.
✓ Add contextual grounding check - Validate if model responses are grounded in the referent source and relevant to the user’s query to filter model hallucination.
✓ Add denied topic - Using a short natural language description, Guardrails helps you to define a set of topics to avoid within the context of your application.
✓ Configure content filters - The safeguard helps filter harmful multimodal content containing topics such as hate speech, insults, sex, violence, and misconduct (including criminal activity), and helps protect against prompt attacks (prompt injection and jailbreak)
✓ Add sensitive information filters - Guardrails helps you detect sensitive content such as personally identifiable information (PII) in user inputs and FM responses.
AWS Bedrock Guardrail
Partner
AWS Bedrock provides a comprehensive solution for securing your LLM applications, including content filtering, PII detection, redaction, and more.
✓ Add contextual grounding check - Validate if model responses are grounded in the referent source and relevant to the user’s query to filter model hallucination.
✓ Add denied topic - Using a short natural language description, Guardrails helps you to define a set of topics to avoid within the context of your application.
✓ Configure content filters - The safeguard helps filter harmful multimodal content containing topics such as hate speech, insults, sex, violence, and misconduct (including criminal activity), and helps protect against prompt attacks (prompt injection and jailbreak)
✓ Add sensitive information filters - Guardrails helps you detect sensitive content such as personally identifiable information (PII) in user inputs and FM responses.
AWS Bedrock Guardrail
Partner
AWS Bedrock provides a comprehensive solution for securing your LLM applications, including content filtering, PII detection, redaction, and more.
✓ Add contextual grounding check - Validate if model responses are grounded in the referent source and relevant to the user’s query to filter model hallucination.
Patronus AI
Partner
Patronus excels in industry-specific guardrails for RAG workflows.
✓ Retrieval Answer Relevance: Checks whether the answer is on-topic to the input question. Does not measure correctness.
✓ Custom Evaluator: Checks against custom criteria, based on Patronus evaluator profile name.
✓ Is Concise: Check that the output is clear and concise.
✓ Is Helpful: Check that the output is helpful in its tone of voice.
✓ Is Polite: Check that the output is polite in conversation.
✓ No Apologies: Check that the output does not contain apologies.
✓ No Gender Bias: Check whether the output contains gender stereotypes. Useful to mitigate PR risk from sexist or gendered model outputs.
✓ No Racial Bias: Check whether the output contains any racial stereotypes or not.
✓ Detect Toxicity: Checks output for abusive and hateful messages.
✓ Detect PII: Checks for personally identifiable information (PII) - this is information that, in conjunction with other data, can identify an individual.
✓ Detect PHI: Checks for protected health information (PHI), defined broadly as any information about an individual’s health status or provision of healthcare.
Patronus AI
Partner
Patronus excels in industry-specific guardrails for RAG workflows.
✓ Retrieval Answer Relevance: Checks whether the answer is on-topic to the input question. Does not measure correctness.
✓ Custom Evaluator: Checks against custom criteria, based on Patronus evaluator profile name.
✓ Is Concise: Check that the output is clear and concise.
✓ Is Helpful: Check that the output is helpful in its tone of voice.
✓ Is Polite: Check that the output is polite in conversation.
✓ No Apologies: Check that the output does not contain apologies.
✓ No Gender Bias: Check whether the output contains gender stereotypes. Useful to mitigate PR risk from sexist or gendered model outputs.
✓ No Racial Bias: Check whether the output contains any racial stereotypes or not.
✓ Detect Toxicity: Checks output for abusive and hateful messages.
✓ Detect PII: Checks for personally identifiable information (PII) - this is information that, in conjunction with other data, can identify an individual.
✓ Detect PHI: Checks for protected health information (PHI), defined broadly as any information about an individual’s health status or provision of healthcare.
Patronus AI
Partner
Patronus excels in industry-specific guardrails for RAG workflows.
✓ Retrieval Answer Relevance: Checks whether the answer is on-topic to the input question. Does not measure correctness.
Pillar Security
Partner
Pillar Security is an all-in-one platform that empowers organizations to monitor, assess risks, and secure their AI activities.
✓ Scan Prompt: Analyses your inputs for prompt injection, PII, Secrets, Toxic Language, and Invisible Character
✓ Scan Response: Analyses your outputs for PII, Secrets, and Toxic Language
Pillar Security
Partner
Pillar Security is an all-in-one platform that empowers organizations to monitor, assess risks, and secure their AI activities.
✓ Scan Prompt: Analyses your inputs for prompt injection, PII, Secrets, Toxic Language, and Invisible Character
✓ Scan Response: Analyses your outputs for PII, Secrets, and Toxic Language
Pillar Security
Partner
Pillar Security is an all-in-one platform that empowers organizations to monitor, assess risks, and secure their AI activities.
✓ Scan Prompt: Analyses your inputs for prompt injection, PII, Secrets, Toxic Language, and Invisible Character
Prompt Security
Partner
Prompt Security detects and protects against prompt injection, sensitive data exposure, and other AI security threats.
✓ Protect Prompt: Protect a user prompt before it is sent to the LLM
✓ Protect Response: Protect an LLM response before it is sent to the user
Guardrails can be configured per route, per model, or applied globally across OpenAI calls.
Prompt Security
Partner
Prompt Security detects and protects against prompt injection, sensitive data exposure, and other AI security threats.
✓ Protect Prompt: Protect a user prompt before it is sent to the LLM
✓ Protect Response: Protect an LLM response before it is sent to the user
Guardrails can be configured per route, per model, or applied globally across OpenAI calls.
Prompt Security
Partner
Prompt Security detects and protects against prompt injection, sensitive data exposure, and other AI security threats.
✓ Protect Prompt: Protect a user prompt before it is sent to the LLM
How to add guardrails to OpenAI with Portkey
Putting Portkey Guardrails in production is just a 4-step process:
Step 3
Enable Guardrail through Configs
Step 4
Attach the Config to a Request
1
Step 1
Create Guardrail Checks
Step 2
Create Guardrail
Actions
Create Guardrail Actions
2
3
4
1
2
3
4
Step 1
Step 1
Create Guardrail Checks
Create Guardrail Checks
Step 2
Step 2
Create Guardrail
Actions
Create Guardrail Actions
Step 3
Step 3
Enable Guardrail through Configs
Enable Guardrail through Configs
Step 4
Step 4
Attach the Config to a Request
Attach the Config to a Request
Guardrail action settings
Guardrail action settings
Async (TRUE)
Run guardrails in parallel to the request.
→ No added latency. Best for logging-only scenarios.
Async (TRUE)
Run guardrails in parallel to the request.
→ No added latency. Best for logging-only scenarios.
Async (TRUE)
Run guardrails in parallel to the request.
→ No added latency. Best for logging-only scenarios.
Async (FALSE)
Run guardrails before request or response.
→ Adds latency. Use when the guardrail result should influence the flow.
Async (FALSE)
Run guardrails before request or response.
→ Adds latency. Use when the guardrail result should influence the flow.
Async (FALSE)
Run guardrails before request or response.
→ Adds latency. Use when the guardrail result should influence the flow.
Deny Request (TRUE)
Block the request or response if any guardrail fails.
→ Use when violations must stop execution.
Deny Request (TRUE)
Block the request or response if any guardrail fails.
→ Use when violations must stop execution.
Deny Request (TRUE)
Block the request or response if any guardrail fails.
→ Use when violations must stop execution.
Deny Request (FALSE)
Allow the request even if the guardrail fails (returns 246 status).
→ Good for observing without blocking.
Deny Request (FALSE)
Allow the request even if the guardrail fails (returns 246 status).
→ Good for observing without blocking.
Deny Request (FALSE)
Allow the request even if the guardrail fails (returns 246 status).
→ Good for observing without blocking.
Send Feedback on Success/Failure
Attach metadata based on guardrail results.
→ Recommended for tracking and evaluation.
Send Feedback on Success/Failure
Attach metadata based on guardrail results.
→ Recommended for tracking and evaluation.
Send Feedback on Success/Failure
Attach metadata based on guardrail results.
→ Recommended for tracking and evaluation.
Build your AI app's
control panel now
Build your AI app's
control panel now
Build your AI app's
control panel now
Manage models, monitor usage, and fine-tune settings—all in one place.
Manage models, monitor usage, and fine-tune
settings—all in one place.
Manage models, monitor usage, and fine-tune settings—all in one place.
Portkey is a comprehensive platform designed to streamline and enhance AI integration for developers and organizations
Portkey, Inc
2261 Market Street #5205,
San Francisco, CA
Products
© 2025 Portkey, Inc. All rights reserved
HIPAA
COMPLIANT
GDPR
Portkey is a comprehensive platform designed to streamline and enhance AI integration for developers and organizations
Portkey, Inc
2261 Market Street #5205,
San Francisco, CA
Products
© 2025 Portkey, Inc. All rights reserved
HIPAA
COMPLIANT
GDPR
Portkey is a comprehensive platform designed to streamline and enhance AI integration for developers and organizations
Portkey, Inc
2261 Market Street #5205,
San Francisco, CA
Products
© 2025 Portkey, Inc. All rights reserved
HIPAA
COMPLIANT
GDPR