The Sether blog / Field notes
PII redaction for AI, in practice.
Practical guides, release notes and honest comparisons from the people building Sether. Keep the useful context. Get more deliberate about the details you share.
Reading with an AI assistant? Every post is available as Markdown. You can also read the blog in llms-full.txt.
SetherThe idea, illustrated
Your map. Your side.
Replace detected details before sending. Restore recognized replacements using the matching retained map.
Fictional example. Detection can miss values; review the edited text.The featured read.
One useful place to start. Then follow the details that matter to your workflow.
A quote in a name could add a bcc to an AI tool call. Here is the fix in Sether 0.8.0.
Sether 0.8.0 fixes JSON restore injection paths. See the affected scope, JSON-aware APIs, strict mode, runnable examples and upgrade limits.
Read the articleNotes worth opening.
Implementation, trade-offs and product updates. The useful bits, with the caveats still attached.
-
Sether™ / Field notes
release
Safe restoration. In Python too.
Restore parsed values before an authorised action.
A quote in a name could add a bcc to an AI tool call. Here is the fix in Sether for Python 0.4.0.
Sether Python 0.4.0 adds JSON-safe restore and strict token handling. Review the fix, runnable examples and structured-stream upgrade limits.
Read the article -
Sether™ / Field notes
pii redaction
A useful task. Fewer exposed details.
A practical guide to the AI privacy boundary.
PII redaction for LLM applications: the complete guide (2026)
Design an LLM redaction flow: detection, local mappings, reply restore, streaming limits and practical acceptance checks for your application.
Read the article -
Sether™ / Field notes
json
Keep the structure. Replace the detail.
Redact supported values in structured data.
How to redact PII in JSON and structured data before sending it to an LLM
Redact supported personal values in JSON, verify syntax and coverage, then restore replies safely with JSON-aware APIs in TypeScript and Python.
Read the article -
Sether™ / Field notes
ai document generation
Write the document with stand-ins.
Restore authorised originals at the output boundary.
Tokenize PII before AI document generation: keep real customer data out of the model
Separate AI drafting from sensitive field insertion. Use typed replacements, validate the draft and restore known mappings at document rendering.
Read the article -
Sether™ / Field notes
typescript
A boundary before the API call.
Connect TypeScript detection, replacement and restore.
How to redact PII before sending prompts to OpenAI (TypeScript)
Integrate Sether in a TypeScript LLM request path. Test detection, vault sharing, streaming boundaries and JSON-aware reply restoration.
Read the article -
Sether™ / Field notes
python
Keep Python prompts deliberate.
Select supported details to replace before the model.
How to redact PII from LLM prompts in Python
Build a Python redact-and-restore flow with Sether. Check vault sharing, supported wrappers, streaming behavior and JSON-safe restoration.
Read the article -
Sether™ / Field notes
gdpr
A control needs a wider plan.
Review data handling alongside technical redaction.
Is it GDPR-compliant to send customer data to OpenAI?
An engineering guide to reviewing LLM data flows: GDPR scope, provider roles, retention, minimisation and evidence for qualified privacy advice.
Read the article -
Sether™ / Field notes
presidio
Choose by the workflow.
Compare recognition, integration and operational needs.
Microsoft Presidio alternatives for Node.js and TypeScript (2026)
Compare Presidio and Sether integration choices: detection fixtures, reversible operations, streaming, state and deployment boundaries.
Read the article -
Sether™ / Field notes
pii redaction
A stand-in goes out. Originals return.
Keep the mapping under application control.
Reversible PII redaction: how redact-then-restore works
How typed tokens and a retained vault support reply restoration, with practical checks for expired mappings, JSON output and tool authorization.
Read the article -
Sether™ / Field notes
shadow ai
Review the draft before Send.
Keep useful context while inspecting sensitive details.
How to stop pasting sensitive data into ChatGPT (without banning it)
Review sensitive drafts with Sether Shield: local text processing, Free essentials, optional Pro workflows and the limits of browser-side controls.
Read the article -
Sether™ / Field notes
streaming
One value can arrive in pieces.
Test the boundary between streamed chunks.
The chunk-boundary bug: why streaming PII redaction is hard
Why per-chunk scanning misses split values, how bounded buffering helps and which encoding, transport and long-input cases to test.
Read the article -
Sether™ / Field notes
ner
Different text. Different evidence.
Compare structured patterns with prose recognition.
Regex vs NER for PII detection: when patterns aren't enough
Evaluate patterns, labels, known terms and NER for PII detection. Build representative fixtures and review local model deployment boundaries.
Read the article -
Sether™ / Field notes
supply chain
Inspect what you ship.
Treat a scanner result as dated evidence.
How I took my npm package's supply-chain score from 78 to zero alerts
A June 2026 Sether hardening account: dependency provenance, scanner limits, SDK wrapper types and practical checks for the release you deploy.
Read the article
Rather see it than read about it?
Explore the illustrated Shield workflow, or try the sandbox with fictional text. Review what is detected before deciding what to share.