EuropeanAI Newsletter #73: UK's AI Regulation
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.
The European Telecommunications Standards Institute (ETSI, via OCG AI working group) appears to have recommended that standardization organizations should be tasked with the technical definition of AI, as well as the description of a categorization framework for high-risk AI in a letter to the European Commission.
Standards and standardization organizations play a key role in light of the EU’s ongoing work on regulation: the AI Act currently states that it is possible for standards, if they exist and match the legal requirements, to replace specific aspects of the conformity assessment on those requirements. Relatedly, the EuropeanAI newsletter recently covered the European Strategy on Standardisation and possible impacts of the EU-US Trade and Tech Council’s workstreams on standardization and risk management for AI. However, even when all of this will be in further motion, it would remain difficult to assess the vast majority of AI systems and algorithms already deployed. To tackle this issue, Stanford's Cyber Policy Center and the Stanford Institute for Human-Centered Artificial Intelligence launched an AI Audit Challenge. The goal is to focus on illegal discrimination against protected categories and develop tools to asses AI systems for this.
This week’s newsletter looks at ongoing EU policy discourse on data governance. In particular, we wanted to explore this topic in light of the first partial compromise text on the Data Act circulated by the Czech Presidency (Council of the EU). The text appears to have made some novel proposals surrounding the definitions and scope, data sharing and business relations. Specifically, it covers an expansion of the definition to include “any item able to collect or generate data and communicate it”, it introduces accessibility rights for data contributed (including metadata), and, similar to some lighter obligations for small and medium sized enterprises under the AI Act, the Data Act excludes micro and small companies from data sharing obligations. On the latter, the Czech text proposes that “the exemption should continue to apply to companies that have reached the medium size by less than one year”. This compromise text will kickstart discussions in the Council’s Telecom working party today.
Policy, Strategy and Regulation
Canada's Artificial Intelligence and Data Act: taking inspiration from the EU?
The Canadian government tabled a draft law on artificial intelligence. The proposal for the Artificial Intelligence and Data Act (the “AIDA”) envisages uniform requirements vis-à-vis private entities for the design, development and deployment of AI systems. Similarly to the AI Act in the EU, the AIDA takes a risk-based approach to regulating AI and targets systems that are conceptualized as “high-impact” (where the EU’s proposal conceptualizes them as “high-risk”). The “high-impact” concept itself is yet to be defined by regulations. Moreover, the AIDA establishes requirements for risk assessment and mitigation, monitoring for compliance with and effectiveness of the mitigation measures, record keeping and notification of material harm. However, unlike the prescriptive approach taken under the EU AI Act, under the AIDA those requirements are succinct and broadly versed. The AIDA also bans conduct in relation to AI systems that may result in serious harm to individuals or harm to their interests.
Notes in the Margins: The similarities between the EU’s AI Act and Canada’s AIDA strongly suggest that the latter may be drawing inspiration from the former with respect to regulatory tools. In doing so, it also indicates that the so-called “Brussels effect'' (cf. Anu Bradford) – i.e. the process of de facto export of EU regulatory approaches to the rest of the world -- may already be happening with respect to AI regulation. See more on this in edition 69 of the EuropeanAI newsletter where we explored the US-EU Trade and Tech Council.
Governing Data and Artificial Intelligence for all
The European Parliament’s Panel for the Future of Science and Technology published a study on the models for sustainable and just governance of data and artificial intelligence. It assesses the EU’s data governance strategy and existing framework, identifying deficiencies that may impact the development and deployment of AI systems for public benefit. Those deficiencies broadly revolve around shortcomings in conceptualizing data as a public good, fragmentation of the EU’s data governance framework, and insufficient checks and balances for the public over data and AI governance. Proposed policy options include: upholding data as a public good, taking a more coherent legislative approach to data and AI governance, increasing access to data and infrastructure for smaller developers and communities, and more involvement of civil society in data and AI governance, as well as oversight.
Ecosystem
The EU's global approach to research and innovation
The European Parliament published a briefing on the European Union’s global approach to research and innovation. The briefing takes stock of the EU’s scientific research cooperation with third countries within the EU’s research and innovation framework programmes and under various multilateral and bilateral initiatives. It highlights the importance of joint research and technological cooperation in the field of AI and other emerging technologies, including through the EU-US Trade and Technology Council. The EU promotes technological cooperation in AI specifically also through the International Outreach for Human-Centric AI, a foreign policy instrument set up and steered by the European Commission in collaboration with the European External Action Services.
UK Perspective: The Future of Intellectual Property Right in the context of text and data mining
The UK government announced its plans to introduce a new copyright and database exception which allows text and data mining for any purpose. This exception would essentially allow for text and data mining for commercial purposes without this being considered as an infringement of an intellectual property right over protected materials (e.g. copyright works or databases). The exception would be introduced with view to supporting AI development and wider innovation in the UK, in line with the UK’s National AI Strategy.
Notes in the Margins: This development can be read also in the broader context of the identified deficiencies in Europe’s intellectual property protection regimes, of which we have reported previously, and the need to overcome them in order to boost AI development in the region.
Enjoy learning about Europe? Share the subscription link with friends, colleagues, enemies...
Contact Charlotte Stix at:
www.charlottestix.com
@charlotte_stix
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.