The EuropeanAI Newsletter
Covering Artificial Intelligence in Europe

Welcome to the EuropeanAI Newsletter covering the European AI and technology ecosystem. If you want to catch up on the archives, they're available here.

You can share the subscription link here

Germany has taken over the G7 presidency and the G7 Leader’s Communiqué (published last week) touched on some AI relevant areas such as standards towards digital transformation, multilateral dialogues (.e.g the Trade and Technology Council or the Global Partnership on AI), and free flow of data.

The Czech presidency of the Council of the European Union appears to have set out
their areas of focus for the AI Act and, last week, MEP Tudorache (LIBE rapporteur on the AI Act) spoke at DIGITALEUROPE’s reception about “the importance of having an innovation-friendly environment for AI”. He has also highlighted a number of amendments he proposed to the AI Act, which would aim to support Europe’s business and innovation environment.

The Guardian published an
op-ed by a former European Commission official and current associate director at the European Policy Centre about opaque lobbying practices from ‘big tech’.

The UK government published
its response to the Medicines and Healthcare products Regulatory Agency’s  ‘Consultation on the future regulation of medical devices in the United Kingdom’ (post-Brexit).

The next deadline for a
connectivity fund from the EU TAILOR network supporting workshops and visits to selected AI labs is the 15th of July. WomeninAI are still taking submissions for their session at the IJCAI-ECAI conference in Vienna (Austria) on ‘Towards Diversity, Equity and Inclusion in AI Research’.

Policy, Strategy and Regulation

Deep Dive: Testing & Sandbox Environments in the AI Act

Last week, Spain announced the launch of a pilot regulatory sandbox for the development and testing of AI systems. The sandbox will provide AI developers with the opportunity to test their AI systems in a controlled environment and under the supervision of the Spanish regulators, who will be working closely with the European Commission.

The possibility for development, training and testing of AI systems in controlled environments under regulatory oversight exists already under the
European Commission’s proposal for the AI Act (Art. 53 and 54). The Commission’s proposal limits testing to testing in a controlled environment. So far, this appears to be also the European Parliament’s preferred approach, given the absence of any specific amendments on those points in the IMCO-LIBE report on the AI Act.

The recently concluded French Presidency of the Council of the European Union has, however, taken a different approach to regulatory sandboxes and testing overall.
The French Presidency’s proposal for amendments to the AI Act would allow for the testing of AI systems in real-word conditions, supervised by the national regulators. The French Presidency’s proposal further envisages the possibility for testing of high-risk AI systems also in real-world conditions outside regulatory sandboxes, subject to a long list of conditions and ongoing regulatory oversight (Art. 54a).

The Spanish sandbox will trial some of those concepts in practice. It will operate in a controlled (as opposed to real-word) environment. Moreover, it will
reportedly operationalize and test the requirements of the AI Act, in an attempt to steer AI development towards safer and responsible outcomes prior to the AI Act’s adoption. The sandbox will be orientated specifically towards SMEs and start-ups, with the view of helping them build compliance capabilities. As part of the work of the sandbox, best practices, guidelines, standards and implementation (including auditing) toolkits for AI development will be drawn up in several iterations. They are expected to inform the public debate around the adoption of the AI Act and standardization efforts at EU level. The Spanish Government plans to make these documents and tools public during the Spanish Presidency of the Council of the European Unions in the second half of 2023.

Testing of AI systems in a real-world environment appears to be an option on the European Commission’s agenda. During the launch of the Spanish sandbox, Commissioner for the Internal Market Thierry Breton
has emphasized that sandboxes complement the establishment of testing and experimentation facilities open to test AI solutions in a real environment. The upcoming Czech Presidency would also support this approach, according to a representative of the Czech government speaking at the launch of the Spanish sandbox. According to Euractiv, testing facilities are also contemplated as a governance tool for supporting member states in the decentralised enforcement of the AI Act.

Within the European Parliament, views appear to vary, judging
by the latest tabled amendments to the AI Act. They include various forms of testing – in a “strictly controlled” environment, in sandboxes in controlled or real-world conditions, and under regulatory supervision or even independently from supervising authorities, and testing in real-world environments under regulatory supervision. Some proposals introduce an obligation for each EU member state to set-up sandboxes. Other proposals advocate for the possibility of sandboxes being set-up also by SMEs, start-ups, companies and innovators.

Note in the Margins: An AI provider’s participation in a sandbox or real-word testing environment would not shield the provider from liability for damages caused by the AI systems (e.g. due to defects, data breaches, incidents) under most of the above legislative proposals. Some proposals by MEPs advocate for leniency from fines for those participants that adhered to the sandbox’s plan, terms and conditions and followed in good faith the guidance given by the regulatory authorities. Such leniency does not, however, appear to dismiss the possibility for damages claims by affected individuals, under the latest tabled amendments to the AI Act. In this sense, the various sandbox proposals and safe testing initiatives should probably be rather viewed – at least for the time being – as efforts to raise compliance awareness among AI providers, to foster cooperation between authorities and the industry, and to build a more resilient European AI ecosystem. Similar endeavours by the European Commission have previously included the setting-up of the Public Sector Open Data for AI and Open Data Platform and the proposed European Digital Infrastructure Consortiums, as well as the funding of the European Digital Innovation Hubs and the European AI-on-Demand Platform.

Enjoy learning about Europe? Share the subscription link with friends, colleagues, enemies...

Contact Charlotte Stix at:

Dessislava Fessenko provided research and editorial support. 
Interesting events, numbers or policy developments that should be included? Send an email!

Disclaimer: this newsletter is personal opinion only and does not represent the opinion of any organisation.

Copyright © Charlotte Stix, All rights reserved.
Twitter Twitter
Website Website
Email Email

This email was sent to <<Email Address>>
why did I get this?    unsubscribe from this list    update subscription preferences
EuropeanAI · 16 Mill Lane · Cambridge, UK CB2 1SB · United Kingdom