# Derivation — eu-2026-1744/art_001/pt_022/d

> COMPOSED TEXT - NOT AUTHENTIC. This is the EU AI Act (Regulation (EU) 2024/1689) with the Digital Omnibus on AI (Regulation (EU) 2026/1744, in force 27 July 2026) applied by a deterministic, gated parser. Only the Official Journal publications are authentic law; no official consolidated version existed when this was generated. Corrigenda R(01)-R(04) do not correct the English text (verified against CELLAR). Every fragment's derivation chain is available in the source graphs (vault: data/graph/). Not legal advice.

- Page: https://eu-ai-act.standards.riskmandate.ai/derivations/art_001__pt_022__d.html
- Structured: https://eu-ai-act.standards.riskmandate.ai/derivations/art_001__pt_022__d.llm.json
- Text version: 2026-07-27 (generated 2026-07-31)
- Site version: v0.1.4
- Provisions root hash: `77c4ca7c208d73adc24b9a41eb29724bee29a41df98b786b85c57852fa3ae6bd`
- Instruction id: `eu-2026-1744/art_001/pt_022/d`
- Operation: replace · level: paragraph
- In force: 2026-07-27
- Payload sha256: `343f660ca2d9a6e8f0956e84358653255d3b15f9b80e2176f971e4ba3554bcbc`
- OJ Formex member: `L_202601744EN.000101.fmx.xml` (CELEX 32026R1744)

## Context (enacting terms)

> Article 57 is amended as follows:

## Instruction (quoted official text)

> paragraph 5 is replaced by the following:

## Quoted payload (official replacement/inserted text)

5.AI regulatory sandboxes established under this Article shall provide for a controlled environment that fosters innovation and facilitates the development, training, testing and validation of innovative AI systems for a limited time before their being placed on the market or put into service pursuant to a specific sandbox plan agreed between the providers or prospective providers and the competent authorities, ensuring that appropriate safeguards are in place. Such sandboxes may include testing in real world conditions supervised therein. Where applicable, the sandbox plan shall incorporate the real-world testing plan referred to in Articles 60 and 60a.;

## Affected provisions

### `eu-2024-1689/art_057/par_005` — amended

**Before (as published):**

5. AI regulatory sandboxes established under paragraph 1 shall provide for a controlled environment that fosters innovation and facilitates the development, training, testing and validation of innovative AI systems for a limited time before their being placed on the market or put into service pursuant to a specific sandbox plan agreed between the providers or prospective providers and the competent authority. Such sandboxes may include testing in real world conditions supervised therein.

**After (composed current text):**

5. AI regulatory sandboxes established under this Article shall provide for a controlled environment that fosters innovation and facilitates the development, training, testing and validation of innovative AI systems for a limited time before their being placed on the market or put into service pursuant to a specific sandbox plan agreed between the providers or prospective providers and the competent authorities, ensuring that appropriate safeguards are in place. Such sandboxes may include testing in real world conditions supervised therein. Where applicable, the sandbox plan shall incorporate the real-world testing plan referred to in Articles 60 and 60a.

- sha256 of composed text: `ac68d95745e8a299a56974faca059e14f7344647de1b1cbb27077d6ad0ad6c2a`
- derivation.json: https://eu-ai-act.standards.riskmandate.ai/provisions/eu-2024-1689/art_057/par_005/derivation.json

## How to check this

Compare the quoted instruction and payload above (hash-anchored to the OJ Formex bytes of CELEX 32026R1744) against the before/after text. That is the whole verification for this provision — about two minutes. Report the result, including the root hash above: https://github.com/Risk-Mandate/Risk-Mandate__EU-AI-Act__Current-Text/issues/new?template=provision-check.yml