Donald Trump’s AI plans face a new test after an election night that changed the balance of power in Washington. The Associated Press reported that Democrats had won control of the US Senate in the 3 November 2026 midterms. CBS News also projected that Democrats would take the House of Representatives, although counting continued in several districts and the AP had not yet called the House when this article was prepared.
That means President Donald Trump’s preferred approach to artificial intelligence could face much closer congressional examination. But a change in the party controlling Congress is not the same thing as a change in AI law. The constructive possibility is that politicians may have to build a more durable agreement on the questions ordinary people care about: scams, children using chatbots, electricity bills and the rules companies must follow.
What happens to Trump’s AI plans now?
Trump has favoured a light-touch national approach that gives American AI developers room to grow and limits conflicting requirements from individual states. In March 2026, the White House published its proposed national AI legislative framework, asking Congress to support innovation while addressing child safety, intellectual property, AI-enabled scams, energy costs and workforce training.
The election changes who would have to negotiate that framework. A Democratic-led Senate can decide which bills receive hearings, what amendments are considered and which nominees advance. If the projected House change holds, Democrats there would also influence legislative priorities and investigations. That could force the administration to seek support outside its own party for the largest changes to federal AI rules.
It does not mean the White House loses control of every AI policy lever. The president retains executive powers and the ability to veto legislation. Existing agency programmes and executive instructions do not disappear because an election produces a different Congress. And Congress must still agree on the actual wording of any new law before it reaches the president’s desk.
The immediate effect may therefore be a change in scrutiny rather than a sudden change in the software people use. The new Congress would normally begin work in January 2027. AI companies, schools and customers should not expect a chatbot’s rules to change overnight because of a projected election result.
A chance to make AI safer without stopping it
There is an important distinction between slowing an industry and asking it to meet clearer standards. A parent worried about an AI companion encouraging harmful behaviour is not necessarily opposed to using AI for homework. A small company concerned about synthetic-voice fraud may still want AI to help with customer service. The question is whether safeguards can make useful systems easier to trust.
Child protection could be a practical starting point. Trump’s own March framework called for parental controls and measures to reduce the risk of sexual exploitation and encouragement of self-harm on services accessible to minors. A new Congress could examine whether those promises need enforceable requirements, independent testing or clearer routes for families to complain. Any such proposal would still need votes and agreement on how it should work.
AI-enabled scams offer another possible area of cooperation. Voice cloning and convincing automated messages can be useful technologies in legitimate settings, but they can also make impersonation harder to spot. Lawmakers could consider duties around reporting fraud, responding to complaints and protecting consumers. Clearer expectations might help trustworthy developers as well as the people targeted by scams. That is a potential benefit, not an outcome delivered by the election itself.
There is a trade-off here. Rules drafted too broadly could make it harder for small developers to compete, or encourage companies to remove useful features simply to avoid legal uncertainty. Weak or inconsistent rules can leave users exposed and businesses unsure what applies. A credible national framework would have to confront both problems instead of treating innovation and protection as opposites.
Electricity bills may be the AI argument everyone understands
For many households, the most concrete AI policy question is not how a model was trained. It is whether building the infrastructure behind it will make local power more expensive. The administration’s own framework says ordinary ratepayers should not be left paying for data centres and proposes ways for facilities to produce power on site.
State-level action shows why this is contentious. California has moved towards more disclosure about the water and electricity used by large computing facilities, as explained in LiveAIWire’s coverage of its data-centre rules. The federal argument is partly about whether national standards should replace or restrict such state measures, and partly about who carries the cost of new infrastructure.
Businesses are also trying to answer that question directly. In a separate look at OpenAI’s planned grid programme, the central issue was whether future facilities could reduce their electricity demand when households need power. Such commitments deserve scrutiny regardless of which party writes the rules. Evidence about actual costs and performance matters more than a reassuring announcement.
Congress could use hearings, spending decisions and legislation to demand clearer answers about ratepayer costs, local resources and the reliability of proposed solutions. Done carefully, that sort of accountability could help AI infrastructure win public confidence. Done badly, it could create delays without resolving the underlying pressure on electricity networks.
The biggest fight could be over state AI laws
Trump’s push for one national framework has a clear attraction for businesses. A company selling the same AI service across the United States would rather follow one understandable set of rules than interpret a patchwork of conflicting obligations. Smaller firms, which cannot afford a large legal department, may have particularly good reasons to want predictable requirements.
Yet state governments have been moving on issues that federal legislation has not settled. A blanket effort to override those rules could remove protections that consumers and communities already rely on. Trump’s December 2025 executive order directed federal officials to challenge certain state AI laws and prepare proposals for a uniform national approach, while identifying areas, including child safety, that the proposed legislation should not automatically override.
A Congress less willing to accept a sweeping federal override may seek narrower wording, stronger exemptions or clearer consumer rights. Equally, disagreements could leave the country with no comprehensive new law. That uncertainty would not be an unqualified victory for AI users or developers. The quality of any compromise would matter far more than the party label on its sponsors.
What to watch after the midterms
The first meaningful signs will come from what the incoming congressional leadership actually puts on the agenda. Hearings on AI scams, protections for children, data-centre bills and state powers would show where scrutiny is moving. Draft legislation, committee negotiations and White House responses would tell us more than election-night slogans about which protections could become law.
Trump can still propose his preferred system, and Democrats cannot simply legislate it away without navigating the presidency and their own internal differences. Equally, a requirement to bargain is not automatically bad for a fast-growing technology. An AI framework that survives serious scrutiny, protects users and remains workable for businesses would be more useful than a rulebook that changes with every political swing.
The midterms have not made artificial intelligence safer. They have changed who gets to question the decisions being made in its name. The positive opportunity is real but conditional: turning an argument over control of AI into practical rules that people can understand, companies can follow and the public can trust.
About the Author
Stuart Kerr is Technology Correspondent at LiveAIWire, covering artificial intelligence, cybersecurity and the social impact of emerging technology. LiveAIWire is an independent, human-led technology publication using AI-assisted research, editorial production and original AI-assisted editorial illustrations under his direction.