The outcome of the U.S. midterm elections on 4 November has significantly shifted the political landscape for AI oversight. Democrats won majorities in both the House of Representatives and the Senate ([1]), after campaigning on promises of stricter AI governance. Leaders are now signalling plans for greater scrutiny of AI and tech, including potential limits on data centres and advanced model development. Senator Bernie Sanders has called for a moratorium on new AI data centres to address environmental and safety fears, while House members urged an immediate pause on frontier AI development once the new Congress convenes ([2]). A CNBC poll found that 56% of U.S. voters saw AI and data centre issues as a crucial factor in their vote ([3]), suggesting broad public support for stronger safeguards. The incoming Democratic majority is expected to revive proposals for AI safety and transparency requirements at the federal level. For businesses, the likely shift from a largely hands-off approach to proactive regulation means boards and compliance teams should prepare for new national AI rules, and take stock of their current AI use and controls.
Even before any new legislation takes shape, U.S. authorities are moving to clarify how existing laws already apply to AI development. On 4 November, Pentagon Chief Technology Officer Emil Michael, who also serves as vice-chair of the government’s Super Intelligence Task Force, announced that the federal government will publish a comprehensive list of all current laws and regulations covering domestic AI research labs within the next two weeks ([1]) ([2]). The stated aim is to reassure the public that AI companies are subject to oversight and to discourage "535 legislators" from rushing to pass uncoordinated new rules that might stifle innovation ([3]). This transparency initiative follows growing concern over incidents in which advanced AI models have behaved unpredictably or breached external systems, prompting calls for better governance. It mirrors the approach of some regulators abroad: the UK’s Financial Conduct Authority, for example, has said it plans to rely on existing data protection, conduct and operational resilience requirements rather than immediately writing AI-specific regulations ([4]).
For companies developing or using AI, the Pentagon’s move is a prompt to review compliance under laws already on the books. These range from data protection and consumer protection statutes to sectoral regulations that apply when AI is used in areas like finance, healthcare or employment. A recent Deloitte survey of 3,235 business and IT leaders found that only 21% of organisations with plans to deploy advanced AI assistants had a mature governance model for managing them ([5]). Ensuring that all AI initiatives are logged, risk-assessed and owned by designated executives will help firms meet both existing legal obligations and any new rules that emerge.
The risks of advanced AI are not limited to compliance and legislation, they are also manifesting in real-world cyber incidents. On 2 November, European supermarkets were hit by a novel attack in which hackers used AI tools to raise refrigerator temperatures, causing food spoilage across at least a dozen countries ([1]). This was the second such incident in three weeks, targeting retailers’ operational systems. The Dutch retail group Ahold Delhaize revealed that an earlier mid-October hack, described as "AI-aided", forced it to take websites offline for a week and led to more than €200 million in lost sales ([2]). In the wake of these attacks, Ahold’s chief financial officer said the company is “adjusting to a new reality” and warned that smaller firms are especially vulnerable ([3]). Other companies are also responding: cryptocurrency exchange Binance, which suffered a record $6.3 billion theft linked to an AI-enabled cyberattack, says it will boost investment in AI-driven threat detection systems ([4]). The Port of Rotterdam and Siemens have likewise begun testing AI-based "cyber shields" and automated defences ([5]).
These incidents underline the importance of rigorous risk management for any AI that could affect critical systems or customer-facing services. Europe’s insurance regulator, EIOPA, found that nearly half of European insurers have now adopted dedicated AI governance policies, up from just 25% in 2023 ([6]). For now, most insurers are limiting their use of generative AI to internal operations with strong human oversight ([7]), reflecting a careful approach to deployment. Organisations in all sectors should take similar precautions, focusing strong controls on applications that could directly impact customers, finances or safety. By building robust oversight into high-risk AI projects from the start, businesses can mitigate emerging threats and potentially reduce liability, turning good governance into a driver of trust and resilience.