EU AI Act deadlines put U.S. firms on notice

The European Union’s AI act rolled out in 2024, calling itself the first-ever comprehensive legal framework on AI worldwide. The act includes a series of staggered compliance deadlines, one of which fell last week, affecting “general purpose AI models with systemic risk.” For US firms that serve EU customers or deploy AI in the EU, this may have significant consequences.

That’s why we asked Center for AI Strategy Advisor and McDermott Will & Schulte attorney Katelyn Ringrose what this means for the comms leaders and their C-Suite execs – including what steps they should take now.

• Industries that rely on advanced, large-scale AI models should be taking notice of the EU AI Act, particularly those deploying models with the potential for widespread societal impact. This is especially pertinent for finance, recruiting, and healthcare sectors where AI-driven decisions could have significant consequences. While the US currently lacks a federal analog to the EU, US state laws designed to ensure fairness, mitigate bias, and promote transparency are quickly filling the gap.
• Companies deploying complex AI systems with the potential to make or inform consequential decisions must address explainability, which requires not only articulating how their systems reach conclusions but also demonstrating how they were designed to mitigate bias and ensure fairness.
• The absence of a comprehensive federal AI law outside of initiatives like the White House’s recently unveiled America’s AI Action Plan has empowered individual states to become leaders in regulation, with states like Colorado, Texas, and California actively passing legislation. These laws address crucial issues ranging from consumer protection to algorithmic bias, creating a robust, albeit varied, regulatory framework that companies must navigate.

Generally, Ringrose advises a proactive approach to AI. “U.S. companies, even those not operating in the European market, should begin designing clear and simple labels that explicitly state when an employee or customer is interacting with an AI system that might be mistaken for human interaction, like a chatbot or automated assistant,” she comments. “For companies, the goal of these transparency measures should not only be to comply with emerging legislation, such as the EU AI Act and certain US state laws, but also to build and maintain user trust.”