OpenAI 개인 정보 보호 파서

OpenAI shipped Privacy Filter — a model that hides PII in text. Defenders use it so data doesn't leak. But every defense has a second side. Attackers need PII too. From logs, dumps, abandoned S3 buckets, stolen inboxes. Same task: find everything personal in a pile of text. Privacy Filter masks. Privacy Parser extracts. Same model, same label taxonomy, same weights. Instead of you get structured spans — what, where, which type. Defense: audit your data before it leaks. Offense: parse someone else's after it did. The tool doesn't know whose side it's on. It's just a good parser. uv venv uv pip install -e ./privacy-filter uv pip install -e ./pii_parser First run downloads the opf 1.5B checkpoint (~3 GB) to ~/.opf/privacy_filter/ . from pii_parser.hybrid import HybridPIIParser parser = HybridPIIParser(device="cpu") result = parser.parse( "Hi Quindle Testwick ([email protected] / +1-415-555-0102), " "account 40702810500001234567, 14 Beautiful Ct, Anytown USA, " "password Priv4cy-Filt3r-2026." ) for s in result.spans: print(f"{s.label:18} {s.text}") private_person Quindle Testwick private_email [email protected] private_phone +1-415-555-0102 account_number 40702810500001234567 private_address 14 Beautiful Ct, Anytown USA secret Priv4cy-Filt3r-2026 python -m pii_parser.cli_model "Alice paid 40702810500001234567 on 2026-05-17." | Backend | Weights | Speed | F1 | |---|---|---|---| PIIParser | none | µs | 1.000 | ModelPIIParser | 1.5B | 500 ms CPU | 0.733 | HybridPIIParser | 1.5B | 600 ms CPU | 0.929 | Hybrid = model + span-merge + regex backstop. Ship this one. opf v2 taxonomy — 8 categories: private_person · private_email · private_phone · private_address · private_url · private_date · account_number · secret text ↓ opf 1.5B → BIOES logits → Viterbi (tuned) → char spans ↓ span-merge (glues Quindle + Testwick) ↓ regex backstop (URL, secret, account — where the model misses) ↓ spans[] python pii_parser/tests/test_hybrid.py Fixture F1: 0.929 Scenarios: 8/8 passed Latency: ~600 ms CPU Apache-2.0.