The 2026 midterm elections are changing the balance of power in Washington. A Democratic majority in Congress is expected to bring closer scrutiny of AI safety, vendor transparency, workers’ rights, and the use of AI in decisions that affect the public. At the same time, divided federal government will make comprehensive legislation harder to achieve. The likely result is not an absence of policy, but a widening gap between the ambition for a unified federal framework and the reality of states and local governments continuing to set their own rules.
For state and local government leaders, the question is not simply which new laws will apply to them. It is how to govern AI systems in an environment where legal obligations, public expectations, and procurement requirements are evolving at different speeds—and sometimes pulling in different directions.
The blue wave’s most consequential effect on public-sector AI adoption may be to strengthen demands for accountability, even where Congress cannot agree on legislation. In that environment, states and local governments will not merely implement federal policy; they will be among the institutions shaping the practical standard for AI adoption in government.
Discussions about the election’s impact often begin with whether Congress will pass legislation. That is an important question, but it is not the only one.
Control of congressional committees gives the majority party the power to set the agenda, call witnesses, request documents, and scrutinize executive branch activity. According to a November 6 Reuters report, Democratic lawmakers plan to advance a range of initiatives addressing AI systems and data centers. At the same time, AI safety organizations are already working to influence who will lead the relevant committees.
It would be a mistake, however, to assume the new Congress will act as a single bloc. Democrats have differing priorities within their own party, and they will have to contend with the White House’s position, Senate rules, and the need for compromise. A Congress that is more active on AI could produce more hearings and proposals, yet relatively little sweeping legislation. Foley & Lardner
For local governments, this is a practical distinction: a change in political tone can affect expectations of vendors and the level of public scrutiny before any binding change in the law takes effect.
One of the central conflicts in AI policy will continue to concern states’ authority to establish their own rules. In 2025 and 2026, the White House advanced efforts intended to limit certain state regulations, including consideration of legal challenges and a recommendation for federal legislation. In July 2025, however, the Senate voted 99–1 against a provision that would have imposed a broad moratorium on state AI regulation. Opposition to that particular proposal was bipartisan, making it difficult to reduce the debate to a simple Democratic-versus-Republican divide. The White House
The election appears to increase the likelihood that this conflict will continue without a sweeping resolution in 2027. The new Congress may be less inclined to advance broad federal preemption, but targeted federal measures, legislative compromises, and legal challenges to particular laws remain possible.
One detail is especially relevant to government leaders: the legislative recommendations published by the White House in March 2026 distinguish between regulation of the private market and rules a state establishes for its own use and procurement of AI. The recommendations seek to preserve states’ ability to govern their activities in those areas. This is a policy proposal, not a guarantee about the language of any future law, but it illustrates why internal AI governance will remain central even if a federal framework advances. The White House
State and local technology leaders therefore need to track two separate questions: what a state may require of companies operating within its borders, and what it requires of agencies and public bodies using AI for their own purposes.
In the debate over AI regulation, it is easy to focus on new laws. But for a company seeking to sell to government, some of the most consequential requirements may appear in procurement documents.
A government does not need to wait for comprehensive federal legislation to ask a vendor what AI capabilities are embedded in its product, how data is handled, what limits apply to its use, or what happens when the system changes. Depending on its procurement authority, it can incorporate such questions into vendor reviews and contract terms.
California already has dedicated processes for procuring generative AI, including privacy assessments and risk-based vendor reviews. In March 2026, the governor signed an order advancing consideration of additional standards and vendor attestations, including in the areas of safety, privacy, and civil rights. Existing procurement requirements should be distinguished from the recommendations the order directed agencies to develop. California Department of General Services
The trend extends beyond one state. In NASCIO’s 2026 survey, 61% of responding state and territory CIOs reported that their jurisdictions had procurement terms or contract provisions for generative AI. This is self-reported evidence that provisions exist, not an independent assessment of their quality or enforcement. NASCIO
The strategic implication is that a public-sector standard for AI can emerge gradually through contracts, even when lawmakers struggle to agree on a single law. If states and local governments ask vendors for similar information, they may influence practices across the market.
At the same time, a proliferation of uncoordinated requirements can create costs and delays. The challenge for procurement leaders in 2027 will be to develop requirements that distinguish among use cases and levels of risk: reviewing a tool that helps draft an internal document need not be identical to reviewing a system that influences eligibility for a public service.
Despite the shift in Washington, many of the questions most consequential to local governments will be decided at the state level. But there is a substantial difference between a law regulating AI developers and a law or directive that applies to a government body using a system.
Colorado, for example, replaced its earlier framework with a new one addressing automated decision-making technologies that influence consequential decisions. According to the state attorney general’s office, the new framework takes effect on January 1, 2027. In Texas, a law that took effect in 2026 includes, among other provisions, certain disclosure obligations for government entities. These are different approaches, with definitions and applicability that must be examined in context. Colorado Attorney General
Even when a law does not apply directly to a city or county, its effects may reach government through vendors. Transparency obligations imposed on system developers, for example, may change the information available to public-sector customers—although governments should not assume that every piece of information they want will be available.
For a CIO or CAIO, this means legal monitoring needs to be connected to an organizational inventory. It is not enough to know that a new law has taken effect; leaders need to know which systems have been purchased, who uses them, for what purpose, and what role the vendor plays in the decision-making process.
State election results may matter here as well. A change in the governor’s office or the composition of a legislature can shift priorities, appointments, and procurement approaches. But there is no reason to assume that every Democratic administration will adopt the same policy or that every Republican administration will oppose regulation. The details of legislation and policy matter more than party affiliation alone.
The public debate over AI now extends well beyond software. The 2026 elections also made data center development a prominent local issue: where facilities will be built, how electricity and water resources will be affected, who will benefit from the revenue, and who will bear the costs.
These are not the same questions that arise when a local government purchases a document-summarization tool. Nevertheless, the two discussions may converge in the same public arena. Residents experiencing the effects of AI infrastructure in their communities may also want to understand how their government uses the technology, which decisions it influences, and what controls are in place. Brookings identifies a range of state and local policy tools that may come under discussion, including permitting, community engagement, and arrangements concerning development and costs. Brookings
For government leaders, public trust is not a substitute for technical controls—and it does not arise automatically from publishing a policy. It depends on the ability to articulate a clear public purpose, explain when a system assists an employee and when it influences a decision, and show residents whom they can contact when a problem arises.
The more consequential a use is for residents, the more important explanation, review, and clear human accountability become. These are not merely communications considerations; they influence how a system should be designed and procured from the outset.
Political debate can create the impression that the central question is whether governments will face more or less regulation. For many, however, the more immediate obstacle is likely to be organizational: how to implement policy consistently when systems, budgets, and responsibilities are distributed across departments.
NASCIO’s 2026 survey illustrates the tension. On one hand, nearly all respondents reported having enterprise policies and procedures for the development and use of generative AI. On the other, only 28% reported dedicated funding for AI initiatives. The survey covers state and territory CIOs and does not directly measure conditions across all local governments; nevertheless, it points to a gap between the technology agenda and the resources allocated to it. NASCIO
That gap matters because AI governance does not end with the initial approval of a tool. Capabilities are updated, vendors change products, employees find new uses, and risk may change as a system moves from one department to another.
Effective governance therefore requires an ongoing cycle: inventory, review, decision, implementation, and monitoring. It also requires clarity about the roles of different functions—technology, cybersecurity, legal, procurement, human resources, and service owners—without assuming that any one department can manage every consequence on its own.
Under these conditions, preparation for 2027 should not begin with an attempt to predict the wording of the next federal law. It should begin with an understanding of the organization’s ability to make decisions and explain them.
First, establish a picture of existing use. An AI inventory should include capabilities embedded in previously purchased products. The goal is to identify consequential uses, the data involved, and the people accountable for them.
Second, connect risk to procurement. Vendor requirements should reflect the context of use, particularly whether a system affects rights, eligibility, employment, or services to residents.
Third, define accountability over time. The decision to adopt a system is a starting point. Governments should determine who reviews product changes, failures, new uses, and complaints.
Fourth, measure both value and risk. AI evaluations should consider the outcome the government seeks—processing time, service quality, or workforce capacity—alongside privacy, security, accessibility, and potential effects on residents.
Fifth, prepare employees and the public for the conversation. Employees need guidance that helps them exercise judgment. Residents need a clear explanation of how the technology is used and who is accountable.
The blue wave is expected to intensify Washington’s debate over AI. It may lead to closer scrutiny of vendors, more legislative proposals, and renewed examination of executive branch policy. But divided government also limits the ability to translate a political agenda into a uniform legislative framework.
States and local governments will continue to operate within that gap. They will set procurement terms, select use cases, define accountability, and answer residents’ questions. Some of those decisions will respond to law; others will be made because government leaders want to know that the systems they rely on deserve public trust.
.png)
.svg)
