Question 1

How does CertaPrompt protect client confidentiality when using AI?

Accepted Answer

CertaPrompt is specifically designed to protect client confidentiality when professionals use AI tools like ChatGPT or Claude. It automatically detects client names, case references, employee records, financial details, and other confidential data, then removes them before your prompt reaches any AI provider. UK and US courts have actively warned about the risks of exposing client data to consumer AI. CertaPrompt ensures you can use AI productively while maintaining client confidentiality, legal professional privilege, and regulatory compliance.

Question 2

What sensitive data does CertaPrompt remove and where does it happen?

Accepted Answer

CertaPrompt detects and removes 55+ types of sensitive data from your prompts, including client names, email addresses, financial figures, account numbers, case references, and confidential business information. This happens entirely on your device using zero-knowledge architecture. Nothing sensitive leaves your browser. You always have full control over what is sent to the AI.

Question 3

How does verification work and how are sources shown?

Accepted Answer

When you receive an AI-generated response, Accuracy Assist verifies claims using proprietary verification technology across 120+ databases, including government registries, corporate filings, and primary sources. Each source is ranked by tier and shown alongside the claim it supports. You receive a confidence score, source-ranked evidence with key quotes, and a clear verdict: Verified, Partially Verified, or Unverified.

Question 4

Which AI tools does CertaPrompt work with?

Accepted Answer

CertaPrompt works with ChatGPT, Claude, Gemini, Mistral, and any other AI tool you use in your browser. It is completely LLM-agnostic. You are never locked into a single provider and can switch models freely.

Question 5

What data is stored and how is my privacy protected?

Accepted Answer

CertaPrompt is built on a privacy-first architecture. Sensitive data removal runs entirely on your device (client-side), so your data never leaves your browser. For verification, claims are checked server-side against our authoritative source network, but you always have the option to strip confidential content first. We never train our systems on your sensitive data. We may use anonymised analytics data, such as feature usage statistics, to improve the service (see our Privacy Notice for full details).

Question 6

How does CertaPrompt support EU AI Act readiness?

Accepted Answer

CertaPrompt generates detailed, timestamped research receipts that record what sources were checked and what verdict was reached. These support your compliance efforts under the EU AI Act (with the August 2026 enforcement deadline) and GDPR by providing an evidential foundation for demonstrating how AI-generated content was checked before reliance. All verification processing runs on EU data centres.

Question 7

How does Research+ identify AI-generated or synthetic companies?

Accepted Answer

As AI makes it easier to generate fake corporate entities, standard web searches are no longer reliable. Research+ bypasses standard search engines and queries our proprietary database alongside Tier 1 government and corporate registries. By evaluating filing histories and primary-source data, it efficiently cuts through synthetic fraud to verify true business entities.

CertaPrompt is undergoing scheduled maintenance.