OpenAI Privacy Filter (opens in new tab)

We've implemented this type of feature years ago. I can make a couple of comments as a result:

1. Sanitising PII data needs to be de-santised on the client in order to keep the UX somewhat functional. For example, if you say my name is John which get's redact to [NAME] and the model responds with Hi [NAME] it needs to be converted back to Hi John. This means that you need to have a mechanism for reversing PII at the layer where the user is interacting. Of course, that is true if your care about user experience.

2. Redacted PII data is practically useless for most intents and purposes. The model wont be able to do much without some data and there are many things that are considered PII. For simple chat system this is fine. For something more complex where the user needs to interact with the LLM this becomes extremely challenging as the LLM may not be able to do anything at all. There is also the chance of hallucination.

Overall, it is a feature that we support at platform level but it is not something people tend to use due to these limitations.

In my mind the only practical thing to do is to remove some types of PII that represent a security risk and make sure that you use a trusted model that purges PII data as quickly as possible. This will require a very different type of system.

Working on this: https://github.com/KevinXuxuxu/anon_proxy, a sort of anonymization proxy to use with LLM providers. It does model (OpenAI privacy filter) + regex PII detection, and replaces them back-and-forth for API requests and responses. With locally hosted detection model, no PII leaves your local environment. I find it very useful especially when you're working on sensitive documents (legal, tax, immigration etc.), hope you find it helpful as well :)

There's some interesting technical details in this release:

> Privacy Filter is a bidirectional token-classification model with span decoding. It begins from an autoregressive pretrained checkpoint and is then adapted into a token classifier over a fixed taxonomy of privacy labels. Instead of generating text token by token, it labels an input sequence in one pass and then decodes coherent spans with a constrained Viterbi procedure.

> The released model has 1.5B total parameters with 50M active parameters.

> [To build it] we converted a pretrained language model into a bidirectional token classifier by replacing the language modeling head with a token-classification head and post-training it with a supervised classification objective.

I'm no where near as smart as OpenAI of course, but I did build https://tools.nicklothian.com/webner/index.html that uses a BERT based named-entity-recognition model running in your browser to do a subset of PII redaction.

It works pretty well for the use cases I was playing with.

The OpenAI model is small enough that I might enhance my tool to use it.

Just so people are clear, these types of models are almost universally naive and basic. If all you have is a single generic neutral message, "Hi, this is Bob.", it will be sufficient in most cases. If you have a pile of data, I am not aware of any PII redaction tool that has factored in all of the risks to identity leakage.

The problem is when companies use things like this and somehow believe they are anonymizing the data. No, you are not.

Still, for scenarios where the processed data isn't being directly published or shared, but used as some intermediate step like moderation enforcement, human evaluation layers or model training it can be useful to filter these things out.

I welcome this release. Lots of good reasons to have these models and practices in place, even outside regulated industries. Even the EU AI act make some of this encessary (in theory). I've built redaction and rehydration through specilized NER models into https://grepture.com, so definitely adding this to the pipeline. Optionally sitting in the hot path, allowing actual tinkering with requests before/after they hit the LLM can be really handy for compliance or dircet user input scenarios.

On a side note, when I click the link it redirects me to machine-translated version of OpenAI website with completely botched meaning - the word “redacted” is translated to a false friend “redagować” which means to edit/refine text, not anonymize.

It would be nice if their examples weren’t mostly things that are easy to catch with regex, but it’s cool to see if released as an open, local model.

Curious how this compares to presidio which mixes regex with a model: https://microsoft.github.io/presidio/

I ran a comparison using this OPF with what I have in clarity.evergreen-labs.org (it's a Tauri app with local PII redaction.)

A few things jump out:

1) Dates are aggressively redacted, creating false positives. 2) Non-English names are not working yet.

Someone has created the reverse of it: https://github.com/chiefautism/privacy-parser

SuperagentLM made available on-edge PPI redaction models already a few years ago in sizes 20B, 3B, 200M. They still seem to be available via their legacy API - well worth checking out to compare against this one. https://docs.superagent.sh/legacy/llms/superagent-lm-redact-...

I'm surprised nobody else has commented on this. This is a very straightforward and useful thing for a small locally runnable model to do.

> The model is available today under the Apache 2.0 license on Hugging Face (opens in a new window) and Github (opens in a new window).

Bringing back the Open to OpenAI..

50M effective parameters is impressively light. Is there a similarly light model on the prompt injection side? Most of the mainstream ones seem heavier

PII masking is always an afterthought when building a backends 50m active params means it can actually run in production.

Can someone explaon how can I reconstruct the original entities back if there are, for example, more than one person names?

I assume they use this model to be able to train new models with user data.

This is exactly where stochastic approaches feel uncomfortable.

For anything touching security or privacy, even small inconsistencies can quickly erode trust.

This is where stochastic approaches start to feel a bit uncomfortable.

Even small mistakes can make something dealing with sensitive data hard to trust. It seems useful as a first pass, but I’d probably still want some deterministic checks or a human in the loop to feel confident using it.

This looks actually useful. But can someone help me understand how you address the non-perfect scores: "Privacy Filter achieves an F1 score of 96% (94.04% precision and 98.04% recall)."

How would you actually use this if it can fail redacting 4% of the data. How do you reliably know which 4% failed?

Where's the gguf from Unsloth and co?