he European Union released its first set of guidance on February 4th defining compliance with the EU AI Act passed March 2024.
-
The AI Act categorized AI systems by risk ranging from minimal risk (unrestricted use), limited risk (require disclosure, among other things), high risk (includes most HR uses of AI and requires significant restrictions on use), and unacceptable risk (prohibited uses).
-
The AI Act’s implementation timeline spans several years with February marking the first deadline for identifying prohibitions on unacceptable risk AI and AI literacy requirements.
What’s prohibited: The non-binding draft guidance issued February 4, 2025, identifies AI practices that are deemed unacceptable, including some that could have workplace implications. The guidance prohibits AI used for:
-
Social or emotional scoring by using AI that classifies people based on social behavior or personal characteristics or that infers emotions at the workplace or in education institutions.
-
Biometric categorization of people based on their biometric data to deduce or infer their race, political opinions, trade union membership, religious or philosophical beliefs, sex-life or sexual orientation.
-
Manipulating behavior through subliminal techniques or exploitation of a person’s vulnerabilities.
The Act’s obligation for “AI literacy” also went into effect requiring those who use AI to have sufficient skills, knowledge and understanding to handle the systems responsibly.
What’s next for the EU AI Act: The next deadline for Act’s implementation is August 2025 when companies must conduct AI audits and identify compliance officers.
In the U.S., the Trump administration took early action by rescinding President’s Biden’s executive order on AI and creating Stargate, a private-sector AI infrastructure program designed to invigorate AI development.
As part of the rescission executive order, Trump called on all “departments and agencies to revise or rescind all policies, directives, regulations, orders, and other actions taken under the Biden AI order that are inconsistent with enhancing America’s leadership in AI.”
Trump issued a second executive order calling for an action plan to “sustain and enhance America’s global AI dominance…” The written action plan is due to the President by the end of July. It is not clear whether the existing AI Institute housed within the National Institute for Standards and Technology (NIST), developed during the Biden administration, will continue.
The bottom line: A lack of U.S. federal regulation increases the relevance of the EU’s AI Act and will encourage additional action on the state level, requiring employers to comply with differing standards while governing their own AI implementation.
