After US v. Heppner, courts are treating public AI prompts as a potential privilege waiver. PromptShielder masks client names, matter details, and financial data in the browser — before anything reaches ChatGPT, Claude, or Gemini.
Written for the managing partner.
In US v. Heppner (S.D.N.Y., Feb 2026), documents created using a public AI platform were held NOT protected by attorney-client privilege. State and city bar associations are now issuing pointed guidance: firms need realistic, defensible AI policies — not blanket bans.
Banning AI doesn't work. Associates use personal ChatGPT on their phones instead, and now client matter details sit in a consumer account you can't audit. PromptShielder gives your team a compliant path: paste the matter, work with the model, and demask locally. Client data never leaves the browser.
Associates pasting matter summaries, client emails, and settlement terms into public LLMs — the exact fact pattern courts are now scrutinizing.
Ban ChatGPT at the firm and it moves to phones. You lose all visibility into what client data leaves the practice.
New bar association guidance expects firms to document how AI handles client information. 'We told them not to' is no longer a defensible answer.
Our frictionless 3-step loop keeps your workflow fast and your data invisible to AI providers.
Paste your raw text into PromptShielder. Our algorithm instantly detects and replaces names, addresses, emails, and identifiers with secure tokens.
[NAME_1]Copy the sanitized text and safely paste it into ChatGPT, Claude, or any public AI model. Run your prompts with absolute peace of mind.
[EMAIL_1]Take the AI's response and paste it back into PromptShielder. Our local browser dictionary immediately swaps the tokens back to your original, real data.
→ real dataSelect the tier that fits your team's operational scale. All plans include 100% client-side privacy guarantees.
Designed for independent freelancers and solo operators.
or $59 one-time