Best PII Redaction API's | Features, Reviews + More


Juan Combariza

Updated March 2024 | 7 min. read


Table of Contents

Top Choices:


Build Pipelines for PII Redaction

Get free credits to run PII Redaction on a live data pipeline

Redacting PII for Data Compliance


A frequent roadblock that comes up during data initiatives is the fear of data compliance penalties. Considering that GDPR penalties are reaching billion-dollar annual figures, it’s understandable that concerns about adhering to stringent laws like HIPAA and CCPA can bring even enthusiastic data projects to a standstill.

Data professionals now face another layer of complexity, as they have to think about legal aspects, consult with privacy professionals, and beef up their cybersecurity infrastructure.

Fortunately, PII redaction tools exist to streamline data compliance and enable companies to utilize data while staying within the confines of stringent legislation.

Why We Wrote This PII Redaction API Guide

At Datastreamer, we’re on a mission to make unstructured data adoption effortless. We chose a PII redaction tool to integrate in our platform so that our users could secure sensitive data flowing through their pipelines. 

We did our homework to compare PII redaction/data masking solutions, and hope our findings give you a head start in your evaluation process.

The Benefits of PII Redaction

Picture a world where sensitive data is never stored to begin with. By actively stripping away PII from data sets before they even enter organizational workflows, the chances of sensitive information being exposed are drastically reduced.

The benefits of PII redaction include:

  • Minimizing the risk of data breaches
  • Enabling data usage while preserving user privacy
  • Streamlining regulatory adherence with automated processes
  • Enhancing consumer trust and reducing reputational risk


Which enables organizations to remain compliant while:

  • Training AI models with human-provided data.
  • Extracting insights from user-generated content (e.g. social media, dark web, forums, and company documents).
  • Sharing data between stakeholders securely.


💡Reducing Integration Costs of PII Redaction

The upfront costs for PII redaction tools usually isn’t a major hurdle. However, for data teams that have to juggle multiple data sources and various AI tools – maintaining a diverse infrastructure can quickly inflate engineering costs.

Enter Datastreamer, a platform that reduces the engineering efforts of unifying unstructured data by 90% compared to a DIY solution. Bring together data sources and instantly deploy AI without the headache of maintaining complex pipelines.

What is PII?


PII stands for Personally Identifiable Information. It refers to any data that could potentially be used to identify a specific individual. PII can come in various forms such as names, social security numbers, date and place of birth, addresses, and phone numbers.

PII redaction (Personally Identifiable Information redaction) refers to the process of identifying and removing or obscuring sensitive personal data from documents, records, or data sets. Redaction can be done manually, using specialized software, or automatically using AI-powered tools.

PII Redaction Software vs. API’s

Redaction software and redaction APIs both serve the same fundamental purpose. However, they differ in their use cases, flexibility, and how they integrate into existing systems.

Redaction Software:

While easy to implement, these tools might not scale well for large organizations that need to process vast amounts of data quickly.

  • Standalone Application: Redaction software is usually a standalone program that you install on your system or access through a web browser.
  • File Types: Redaction software generally supports file types like PDFs, Word documents, images and videos.
Redaction APIs:

While APIs may require some technical knowledge, they are built to handle automated, large-scale operations. Many APIs now include visual interfaces to simplify ease of use.

  • Integration: APIs can be used for internal use cases but are designed to be integrated into other software systems, allowing developers to add redaction capabilities to their own applications.

  • Data Types: APIs are typically more flexible in terms of the types of data they can handle, including text streams, database entries, and even real-time data flows.


Deep Dive: Top PII Redaction Tools Comparison

1. Private AI

🔌 Private AI is available within Datastreamer. Start for Free –>


Image from Private AI’s web demo

Founded by privacy and machine learning experts, Private AI’s  mission is to create a privacy layer for software and enhance compliance with current regulations such as the GDPR. Utilizing state-of-the-art transformer architectures, Private AI delivers exceptional accuracy right from the get-go. Their terms and conditions clearly state that you maintain complete ownership of your data, offering an added layer of assurance that your information is secure.

Main Use Cases:

Teams invested in AI model training or data insight extraction, while also needing to comply with standards like HIPAA, GDPR, and CCPA, will find this solution invaluable. It’s particularly relevant for sectors that manage sensitive data, such as healthcare, finance, as well as those dealing with internal corporate documents and user-generated content.

Key Features:

  • Developed by privacy experts, this tool is singularly focused on PII redaction, delivering greater accuracy than other generalized solutions on the market.
  • Recognizes 50+ entities of personally identifiable information in 49 languages.
  • For text: Redact, anonymize, create synthetic PII, or tokenize PII (reversible).
  • For files: Redact PDFs, blur faces in images, bleep out PII in audio, and more. 
  • Native functionality for LLM’s and ChatGPT-like applications.


Data Types Supported:

  • Text, Files (PDF’s), Images, and Audio.
  • Real-time flows and batch data sets supported.
  • Tip: Datastreamer’s managed connectors streamline the integration of data sources into Private AI (chat transcripts, web data, social media, + more).


Pricing Overview:

  • To obtain a pricing quote, you’ll need to reach out to Private AI.
  • Free trial: You can request a free API key for 500 API calls, no payment details required.

Customer Review Highlight:

Gener C.
R2R Process Specialist

“Best data security out there – by employing techniques such as federated learning or differential privacy, it allows organizations to train AI models without directly accessing or exposing individual data.” 

💡Private AI is Available within Datastreamer

Private AI can be deployed instantly through Datastreamer. Our platform saves you 3-6 months of engineering time by automating steps in the procurement, ingestion, and unification of diverse data sources.

2. Assembly AI

AssemblyAI specializes in empowering developers to build LLMs and AI applications that utilize voice data. With a product suite that places a strong emphasis on speech-to-text conversion, their Audio Intelligence capabilities include identifying and redacting both PII and hate speech for content moderation purposes. Before the transcribed text is sent back to you, it can be cleansed of sensitive information like phone numbers and social security numbers.

Main Uses Cases:

AssemblyAI is tailored for teams that want to create AI applications with voice data. Their PII services focus solely on text that originates from audio transcriptions. This makes it a compelling option for teams who want to focus on unique audio formats like virtual conference calls, television audio, or podcasts.

Key Features:

  • Identify and remove PII from text that has been transcribed from audio.
  • Audio PII Redaction: Bleep out instances of PII or other recognized entities.
  • Detect and moderate other entities such as hate speech, violence, sensitive social issues, alcohol, drugs, and more.
  • Connect voice data to LLM’s easily with their LeMUR feature.


Data Types Supported

  • Audio files and video audio files.
  • Transcribed text from audio.
  • * Raw text (non-transcribed) is on the roadmap but is not currently available.
  • * Files (PDF’s, Word) are on the roadmap but are not currently available.


Pricing Overview:

  • To run PII redaction, you must first transcribe text with the their “Core Intelligence” Feature.
  • The core transcription model starts at a base rate of $0.65 per hour, while adding PII redaction comes with extra costs that vary depending on the volume
  • Free Trial: AssemblyAI offers a trial with 5 hours worth of credits included.

Customer Review Highlight:

Nico R.

“I’ve tested many speech to text api’s to transcribe audio/video files and AssemblyAI consistently wins on accuracy (lowest WER on our audio files). Accurate transcription of phone calls and recordings and video interviews.”

3. Amazon Comprehend

Amazon Comprehend offers capabilities in natural language processing including the detection and redaction of Personally Identifiable Information (PII). As part of the AWS family, it carries the recognized brand name. While PII redaction is included as a feature, the platform’s main emphasis is on providing users with the tools to train their own natural language processing models. User reviews highlight issues with accuracy, seemingly due to a lack of native specialization in the pre-built models.

Main Use Cases:

For teams interested in simple PII redaction across text data sources on a smaller scale, this tool has you covered. It supports real-time data streaming and provides smooth integration with AWS services like S3 and Redshift.

Key Features:

  • PII redaction is offered as a minor feature within a broader suite of NLP capabilities.
  • Emphasis on empowering users to train their own models without prior ML experience.
  • Detect and redact instances of PII in text and documents with support for multiple languages.
  • Index and search the data that you have processed. 


Data Types:

  • Text sources: customer support tickets, product reviews, emails, social media feeds, and more.   
  • Text documents: PDF, Word, and more.
  • Audio & Video files require another solution such as Amazon Transcribe.


Pricing Overview:

  • The pricing model can get confusing as it is based on units (character counts) and varies per volume. View the pricing table here. 
  • User feedback indicates that the expenses can get unjustifiably high. It is cost-effective to start, but expensive to scale.
  • Free Trial: Amazon offers a free plan with 50k units worth of text of credits.

Customer Review Highlight:

John C.
Lead Application Developer

“It was effortless to use and set up in our codebase. We used it to remove PII from sensitive client content. We set up the service for use, and added it to our codebase in less than a day. The cost to simply remove PII isn’t scalable, depending on what kind of content you are trying to process.”

4. Microsoft Presidio (Open-Source PII Redaction Tool)

Alongside its commercial cognitive search capabilities, Microsoft offers an open-source PII detection tool for Azure named Presidio. This tool gives developers and data scientists the flexibility to either modify existing PII recognizers or add new ones via API or coding, tailoring the tool to your specific anonymization requirements.

Being open-source and transparent, this tool is an excellent option for smaller research groups or teams that are new to the world of PII redaction.

However, the flip side of opting for a less mainstream open-source solution is that you may encounter limited support options, fewer built-in features, and demand greater technical resources to implement and sustain the system.

Learn more about Presidio in this medium article: Detecting PII With Microsoft Presidio

Link to the GitHub: Microsoft Presidio

5. Microsoft Azure

Microsoft Azure’s AI Services include functionalities that allow for PII redaction in unstructured text through the use of pre-defined entities. Furthermore, its Cognitive Services suite has a specialized “skill” for PII Redaction that works not just on text but also on audio, images, and video content. While these services are native to the Azure platform, it offers the flexibility to be run anywhere via Docker containers.

Learn more in the Microsoft Azure documentation.

6. Super AI

Super AI serves as an intelligent document processing platform with PII redaction tailored for files (documents, invoices, and PDFs, etc.). Safeguard sensitive information through its automated redaction features. The “super redact” function is multilingual and adept at identifying various forms of personal data like phone numbers and social security numbers. Additionally, it provides both reversible and irreversible options for PII redaction.


Build Pipelines for PII Redaction

Get free credits to run PII Redaction on a live data pipeline