OpenAI releases Privacy Filter model to strip sensitive data like passwords and addresses locally before ChatGPT use, boosting on-device AI privacy protection


3 recorded changes
Want your article here?
Promote with Leviathan News

3 recorded changes
Want your article here?
Promote with Leviathan NewsOn-device PII filtering is the cheapest regulatory moat on offer — cheaper than actual E2E, cheaper than TEE attestation, cheaper than Apple's Private Cloud Compute. Filter weights become a single point of failure: miss one pattern and the raw PII hits OpenAI servers and stays there forever. Nillion's blind compute and Oasis Sapphire already ship data-never-decrypts workflows; OpenAI just shipped "trust us, we scrubbed it first" and called it privacy.
Top comment by @Benthic

Reuters ·

Coindesk ·

CNBC ·

Hut8 ·

𝕏/@OpenGradient ·

𝕏/@a16zcrypto ·

Reuters ·

Coindesk ·

CNBC ·

Hut8 ·

𝕏/@OpenGradient ·

𝕏/@a16zcrypto ·
🚀 Love DeFi? Ready to dive in and start earning $SQUID while making an impact?