AI View February 2024

EU AI Act resources

The latest post in our series on the EU AI Act covers the geographic scope of the Act and is available here. Stay tuned for more posts diving into the detail of the EU AI Act over the next month.

This edition brings you:

• EU AI Act passes the latest stage in the EU legislative process
• Department of Science, Innovation and Technology publishes its ‘Introduction to AI assurance’
• The Commission Decision establishing the European AI Office enters into force
• Japan’s Liberal Democratic Party proposes legislation for regulating generative AI
• The Federal Trade Commission publishes supplemental notice of proposed rulemaking to prohibit AI impersonation of individuals
• India to develop draft legislative framework for AI

EU AI Act passes the latest stage in the EU legislative process

On 13 February, MEPs on the Internal Market and Consumer Protection (IMCO) committee and Civil Liberties, Justice and Home Affairs (LIBE) committee in the European Parliament voted to adopt the EU AI Act.

As the next step, a vote for all MEPs is due to take place in the European Parliament on 10 and 11 April. This is the crucial vote, after which the EU AI Act will be published in the Official Journal and then become law 20 days after publication.

See more here.

Department of Science, Innovation and Technology publishes its ‘Introduction to AI assurance’

On 12 February, the UK Government Department of Science, Innovation and Technology (DSIT) published new guidance on AI assurance, which aims at providing a grounding in AI assurance for those less familiar with the technology.

As an introductory guide, rather than getting into the technical detail, the document focuses on the underlying concepts of AI assurance, setting out an accessible framework for understanding AI assurance mechanisms and global technical standards.

The guidance covers:

• AI assurance in context: providing an introduction to the background and conceptual ‘underpinnings’ of AI assurance.
• The AI assurance toolkit: an introduction to key AI assurance concepts and stakeholders.
• AI assurance in practice: giving an overview of different assurance techniques and how to implement AI assurance within organisations.
• Key actions: a brief overview of key actions that can be taken by organisations looking to embed AI assurance.

DSIT has described the guide as a ‘living, breathing document’ that will be updated over time. The ‘Introduction to AI assurance’ is the first in a series of guidance aimed at helping organisations upskill on topics around AI assurance and governance.

Read the introductory guidance here.

Commission Decision establishing the European AI Office enters into force

On 21 February, a Commission Decision of 24 January 2024 establishing the European Artificial Intelligence Office entered into force.

The European AI Office has been established to assist in the implementation of the forthcoming EU AI Act, and is set to perform a number of official tasks (as set out in Article 3 of the Decision) but only in relation to general-purpose AI models and systems under the Act (i.e. not in relation to prohibited AI, high-risk AI etc.), including:

• monitoring the implementation and application of rules,
• monitoring the emergence of unforeseen risks, and
• investigating possible infringements of rules.

Read the full Commission Decision here.

Japan’s Liberal Democratic Party proposes legislation for regulating generative AI

On 16 February, Japan’s Liberal Democratic Party (LDP) published a draft proposal for the regulation of generative AI technologies.

Key highlights include:

• The scope of the legislation is limited to “Specified AI Platform Models” which have a significant social impact and is not intended to target smaller scale AI models or start-up business initiatives.
• Organisations developing these models will be required to establish a system covering seven key areas, including third-party vulnerability verification and publishing of basic model specifications.
• Developers of these models will be required to monitor compliance in the development of systems and periodically report to the government (or the recently launched AI Safety Institute).

Japanese-language announcement and draft proposal available here.

The Federal Trade Commission publishes supplemental notice of proposed rulemaking to prohibit AI impersonation of individuals

On 15 February, the Federal Trade Commission (FTC) published a supplemental notice of proposed rulemaking to prohibit the impersonation of individuals.

According to the FTC, this follows a surge in complaints around impersonation fraud, with emerging technologies such as AI threatening to ‘turbocharge’ this growing issue.

The FTC is seeking public comment on the notice, including on whether it should be unlawful for a firm to provide goods or services (i.e., an AI platform producing images, video or text) that it knows (or should have reason to know) are being used to harm consumers through impersonation.

The supplemental notice is expected to appear in the Federal Register shortly, and once published in the Register the public comment period will be open for 60 days.

Read more here.

India to develop draft legislative framework for AI

On 20 February, the Indian Minister of State for Electronics and Information Technology, Rajeev Chandrasekhar, announced that the government will soon publish a draft regulatory framework for AI.

Set to be released in June or July of this year, the Minister said that the intention of the draft legislation is to harness AI for economic growth and address potential risks and harms.

This statement builds on an early remark from Mr Chandrasekhar that the Indian Government preferred to regulate AI via clear standards to address concerns such as bias as opposed to regulating systems at certain stages of development.

Read more here.