AI risk assessment — run one now, free

An AI risk assessment identifies and scores the risks an AI system poses — prohibited practices, high-risk classification under the EU AI Act, data-protection impacts and required safeguards. This tool runs a real assessment in your browser: answer the questions that apply, and connected modules (DPIA, transfer, vendor) activate automatically as your answers require them.

AI Risk Assessment (EU AI Act)
Free, no signup. Your answers stay in this browser — nothing is stored on our servers unless you choose to create an account.

How it works

  1. Answer the triage questions. The AI-system context is preset; personal data, transfers and vendors activate their own modules automatically — one intake, every connected obligation.
  2. Work through the activated modules. Prohibited practices, high-risk classification, GPAI transparency — with plain-language guidance and optional AI assists on the drafting-heavy questions.
  3. Get scored results instantly. A 0–100 risk score with per-factor reasoning and proposed actions, exportable as PDF, Word, Markdown or JSON.

Frequently asked questions

Do I need an EU AI Act risk assessment?

If your organisation develops, deploys, imports or distributes an AI system that touches the EU market or EU/EEA individuals, the EU AI Act applies on a risk basis. Prohibited practices are already enforceable; high-risk obligations phase in through December 2027. This assessment screens your system against those categories and shows which obligations follow.

Is this tool really free — what's the catch?

The assessment runs entirely in your browser and the full results are shown on-page, free, with no login or email required. Your answers are stored only in this browser. A free account adds saving, sharing and clean exports; free exports carry a small "Prepared with Raksa" mark.

What happens to the answers I enter?

Nothing leaves your browser unless you choose it. Answers autosave to this browser's local storage. If you use an optional AI assist, the relevant answers are sent to generate that one suggestion and are not stored on our servers or used for training.

How is this different from asking ChatGPT to assess my AI system?

A risk assessment is an accountability document, not a chat transcript. This tool runs a structured, versionable assessment with per-factor scoring and proposed actions — and because it shares one nucleus with DPIA, transfer and vendor assessments, connected obligations activate automatically instead of being missed.

One nucleus, every assessment

The same engine powers DPIA screening and transfer impact assessments — arriving on this page shortly. Notice how answering "yes" to personal data above already grew a DPIA section inside this run: that is the nucleus at work. See all assessments.