EU AI Act definition of AI aligns with OECD definition, biometric risk updated
The European Union Artificial Intelligence Act could be taking a significant stride forward as the European Parliament agrees on what AI actually is. Euractiv, which provides excellent coverage of the EU, has seen the text of a political agreement on the definition of AI which is effectively that of the Organization for Economic Cooperation and Development (OECD).
They also report that there are changes to the definition of “significant risk,” biometric authentication and identification. A right to explanation for AI-based decisions could also be incorporated, where individuals can seek an explanation for how an AI-powered decision on them, such as turning them down for a loan, was made.
As the AI Act is intended to prevent harm from AI, the definition of what AI is proves fundamental.
Text seen by Euractiv reads:
“‘Artificial intelligence system’ (AI system) means a machine-based system that is designed to operate with varying levels of autonomy and that can, for explicit or implicit objectives, generate output such as predictions, recommendations, or decisions influencing physical or virtual environments.”
“The OECD defines an Artificial Intelligence (AI) System as a machine-based system that can, for a given set of human-defined objectives, make predictions, recommendations, or decisions influencing real or virtual environments.”
The definition is expected to change further with the removal of “machine-based,” according to Euractiv sources. The similarity to the OECD definition is acknowledged with an addition to the Act’s preamble “to ensure legal certainty, harmonisation and wide acceptance,” notes Euractiv.
The narrower definition of AI is more in line with conservative political groupings as left-of-center politicians had been pushing for a broader, more encompassing understanding of the technology.
Whereabouts the definition appears within the Act has also changed in previous debates on the text. It was moved from Annex I to its own article, preventing the European Commission from subsequently altering the definition, Euractiv reported in November 2022.
Biometric authentication, categorization updates
Compromise amendments were made to definitions of biometric authentication, verification and categorization at a technical meeting on other definitions in the proposed Act on Monday, reports Euractiv, as was the definition for the term “significant risk.”
According to the altered text seen by Euractiv, remote biometric verification systems are defined as comparing biometric data against a database with a person’s prior consent and biometric authentication is when the person asks to be authenticated.
Biometric categorization is on the list of prohibited AI use cases. A further point was added to include the inferring of personal attributes, with the examples of health and gender provided.
The changes, if they become official, are the most recent in a long line of developments and negotiations. The previous set of amendments, in early February, centered on determining high-risk uses.
AI | AI Act | biometric authentication | biometric identification | biometrics | EU | legislation