Philter
philterd.ai
- Free plan
- Windows
- Linux
- API
- Self-hosted
What Philter is
Philter is open-source software for detecting and redacting personally identifiable information and protected health information in text. For detected entities, users can choose to mask or encrypt values, replace them with synthetic ones, remove them, or leave them in place. Policies, custom dictionaries, and identifier patterns can be set up for cases such as medical record numbers and transaction IDs. Its General Purpose, Healthcare, and COVID-19 NLP lenses are intended for different kinds of text. Philter provides an HTTP REST API, a Java SDK, and an OpenAPI specification for generating clients in other languages. It can run in a customer-controlled environment, including firewalled or air-gapped setups; the maker says text is not sent to third-party services. Redaction requests are logged in a tamper-evident, hash-chained audit record that can be queried and exported to CSV. Deployment is available privately or through AWS, Google Cloud, and Microsoft Azure marketplaces. Detection is probabilistic, so results should be checked against an organization’s own data before production use. The open-source toolkit has community support; commercial licenses include direct support.
Who it suits
Philter may suit teams in healthcare, finance, legal, government, or AI workloads that need redaction in APIs or data pipelines. It is also relevant to organizations seeking to keep text processing in their own environment.
Good to know, and what to watch
- Offers several actions for detected sensitive text.
- Supports custom policies, dictionaries, and identifier patterns.
- Can run in firewalled or air-gapped environments.
- Provides a queryable audit log with CSV export.
- REST API, Java SDK, and OpenAPI specification are available.
- Detection is probabilistic and needs validation against your data.
- Open-source toolkit support is through documentation and GitHub issues.
- Commercial use of Philter Desktop costs $100 per user per year.
Plans
Every plan and priceOpen source toolkit
Philter Desktop — Personal & evaluation
Philter Desktop — Commercial
Free plan
Getting started
- Open the Philter website
- Choose the open source toolkit or a desktop plan
- Deploy it on Linux, Windows, an API, or a self-hosted environment
- Configure policies, dictionaries, identifier patterns, and an NLP lens
- Connect through the REST API, Java SDK, or generated OpenAPI client
- Process supported documents or text batches
Limits to check
The Personal and evaluation desktop plan is for individual, non-commercial, and evaluation use. The open source toolkit has no usage caps but does not include commercial review; commercial desktop access is billed per user annually.
The verdict
Philter offers configurable text redaction and deployment options for teams that need control over where data is processed. Treat its probabilistic detection as a reason to validate results before relying on it in production.
Questions people ask
Is Philter open source?
Yes. The open source toolkit is released under Apache 2.0, with source code available on GitHub.
What does the free option include?
Philter Desktop Personal and evaluation costs nothing and is for individual, non-commercial, and evaluation use, with full on-device redaction and GitHub community support. The open source toolkit also costs nothing and is self-hosted.
How much is commercial desktop use?
Philter Desktop Commercial costs $100 per user per year and includes direct creator support, updates, maintenance releases, and upgrades during the subscription.
Where can Philter run?
It supports API, Linux, self-hosted, and Windows deployments, plus private environments and AWS, Google Cloud, and Microsoft Azure marketplaces.
What file types are supported?
Supported inputs include PDF, DOCX, XLSX, CSV, TXT, RTF, EML, and MSG.
Does Philter send text to third parties?
The maker says processing occurs in the customer’s environment, including firewalled or air-gapped setups, without transmitting text to third-party services.
What the maker publishes
- Accuracy limitation
- Detection is probabilistic, and the maker advises validating results against a customer’s own data before relying on them in production. source
- API and SDKs
- Philter provides an HTTP REST API, a Java SDK, and an OpenAPI specification for generating clients in other languages. source
- Audit trail
- Philter records redaction requests in a tamper-evident, hash-chained audit log in the customer’s datastore, which can be queried and exported to CSV. source
- Cloud deployment
- Philter can be deployed in a private environment or through AWS, Google Cloud, and Microsoft Azure marketplaces. source
- Company history
- Philterd says it was founded in 2017 by Jeff Zemerick. source
- Data handling
- The maker says Philter runs in the customer’s environment and does not transmit text to third-party services; it can run in a firewalled or air-gapped environment. source
- Detection and actions
- It detects many entity types and lets users choose to mask, encrypt, replace with synthetic values, drop, or pass them through. source
- Founded
- 2017 source
- Intended users
- The maker describes Philter for healthcare, finance, legal, government, and AI workloads, including teams adding redaction to APIs and data pipelines. source
- Models
- Philter offers General Purpose, Healthcare, and COVID-19 NLP lenses for different text types. source
- Open source and security
- The maker says its products are released under Apache 2.0, with source code publicly available on GitHub for inspection and self-building. source
- Policies and custom entities
- Users can define policies, custom dictionaries, and identifier patterns such as medical record numbers and transaction IDs. source
- Support
- The open source toolkit includes community support through documentation and GitHub issue trackers, while commercial licenses include direct support. source
- What it does
- Philter is open source software that redacts personally identifiable and protected health information from text. source
Checked against the maker's pages on 2 Oct 2026.