Across the economy, AI systems are enabling new forms of fraud, expanding surveillance-based pricing, and increasing costs for essential services such as health insurance, transportation, and electricity.
By: Rishi Bharwani, Janelle Jones, Lindsay Owens, Zoe Butler
Families across the country are already struggling with a cost-of-living crisis. Housing, food, health care, insurance, and utilities have all grown more expensive in recent years. At the same time, artificial intelligence (AI) is a powerful new economic force that is rapidly shaping markets, workplaces, and everyday life.
AI is often promoted as a technology that will boost productivity, reduce costs, and improve efficiency. In theory, those gains could benefit consumers through lower prices and better services. In practice, however, many early applications of AI are shifting costs onto households while concentrating benefits among technology companies and large corporations.
Across the economy, AI systems are enabling new forms of fraud, expanding surveillance-based pricing, and increasing costs for essential services such as health insurance, transportation, and electricity. From higher energy bills driven by data centers to algorithmic pricing that raises travel costs, the economic footprint of AI is increasingly hitting family budgets.
Without effective oversight and policy safeguards, consumers will increasingly be left at the mercy of opaque algorithms and powerful technology firms. This report examines several of the most immediate ways AI is affecting household finances across the U.S.
AI is accelerating the scale and sophistication of online fraud, enabling scammers to use voice cloning and deepfake technology to impersonate trusted people and create other convincing videos or images. About three in four adults have experienced an online scam or attack, and a majority report getting scam phone calls, emails, or texts at least once a week. Federal data show reported fraud losses rose again in 2025 as fraud schemes became more sophisticated and more difficult to detect.
The resulting losses can place severe pressure on household finances. Internet-based scams and cybercrime cost Americans $148 billion in 2025, including $26 billion in California, $13 billion in Texas, and $11 billion in Florida. The true financial burden is likely substantially higher than enforcement data reflects because for every reported crime, six go unreported. The estimate also excludes indirect legal and administrative costs. Victims may exhaust their savings, take on debt, miss work, or pay additional costs to recover compromised accounts and identities. As household budgets tighten due to rising costs for housing, food, and utilities, families cannot afford to lose hundreds, or even thousands, of dollars to sophisticated AI-enabled scams.
While mental health struggles are deeply personal, they are also an economic issue. The annual economic burden of mental illness rivals that of an average recession, an estimated $282 billion per year. For households, those costs include therapy, medication, counseling, lost work, and caregiving. Families with children are particularly burdened: In 2022, out-of-pocket spending on pediatric behavioral health reached $2.9 billion, an increase of about 38% over the previous 11 years. As AI-driven platforms intensify mental health risks, they may also leave families paying more for therapy, medication, and other forms of care.
AI is powering more and more of what people see on social media, and the evidence links heavy use to worse mental health outcomes. Researchers have long linked social media to anxiety, depression, loneliness, and body-image concerns. Roughly 250 million people – or three out of every four people in the United States – actively use social media, often turning to these platforms throughout the day for news, entertainment, and connection. AI now shapes much of what users encounter on social media feeds: On LinkedIn, an AI-detection firm classified 81% of a sample of long-form posts as likely AI-generated. Meta told investors in July 2026 that every public Reels and Feed post on Instagram is processed and analyzed by an LLM, with the results feeding ranking and recommendations.
AI-powered recommendation systems use behavioral data, including viewing time, searches, clicks, and reactions, to personalize users’ feeds. The systems continually learn from each interaction, which means that engaging with distressing content can lead to greater exposure to similar material. These feedback loops may reinforce compulsive use, unhealthy social comparison, and patterns of content consumption associated with anxiety, depression, and body-image harms.
Growing evidence links the patterns encouraged by AI-powered social media platforms with declining mental health outcomes. Frequent AI interaction and distress from coping with technology are associated with higher rates of anxiety and depression. One study found that adults who use AI daily are 30% more likely to experience at least moderate depression than adults who do not use AI at all. Young people are especially vulnerable. More than a third of U.S. teenagers say they use at least one major social media platform almost constantly, and nearly half say it is mostly harmful to people their age – up from 32% in 2022. In March 2026, a Los Angeles jury in the first major trial against Meta and YouTube found that their platforms were designed in ways that harmed a young user’s mental health.
Generative AI chatbots create a separate set of mental health risks, especially for young people. Unlike recommendation systems that shape what users see, chatbots communicate directly with users and may be treated as sources of emotional support or mental health guidance. Nearly one in five youth ages 12 to 21 report using generative AI for mental health advice, raising concerns about inaccurate guidance, emotional dependency, and delayed access to professional care.
These trends illustrate how the expansion of chatbots and AI-driven social media may be amplifying both the prevalence and cost of mental health challenges. Families already spend hundreds or even thousands of dollars each year addressing mental health needs, and the growing influence of AI-based systems may further intensify these financial pressures. Whether through stress from frequent technology use, harmful advice, or the promotion of unhealthy comparisons, AI-powered media platforms are increasingly affecting mental well-being, household finances, and broader economic outcomes.
AI is changing the way patients experience the health care system and driving up costs. Health care expenses are already one of the largest budget lines for families. In 2025, workers paid an average of $6,850 toward annual premiums for employer-sponsored family coverage, and the total family premium rose 6% to nearly $27,000 – up more than 50% over the past decade. For those who purchase coverage on the Affordable Care Act Marketplace, the premiums enrollees actually pay rose an average of 58% this year following the expiration of enhanced subsidies. Federal policy changes are often cited as the primary explanation for these increases. However, the rapid adoption of AI across the health care industry is another powerful and less visible driver of higher costs.
Most major health insurers now use AI to review claims, process prior authorizations, and identify fraud. UnitedHealthcare, the nation’s largest health insurer, projects nearly $1 billion in operating cost reductions in 2026, much of it AI-enabled, but those savings are not being passed on to patients as lower costs.
One example is prior authorization. AI systems can rapidly review claims and flag services for denial, but physicians increasingly report that these systems are creating new barriers to care. In the American Medical Association’s latest survey, 60% of physicians said they are concerned that AI is increasing prior authorization denials. AI adjudication tools like UnitedHealth’s nH Predict have been accused of often erroneously denying needed care, hoping that patients and providers don’t take the time to appeal. A recent lawsuit alleges that the 0.2% of patients who do appeal win about nine times out of ten, reversing the AI-powered tool’s denial.
Since January 2026, Medicare has been piloting an AI system to review prior authorization in six states. If AI prior authorization systems deny more care, insurance companies pay out fewer claims, which could decrease premiums. However, these projected savings may be offset by the costs of denials and appeals. When claims are denied more often, providers must devote more staff time and resources to filing appeals, and patients bear the burden in both time and money. Denials also don’t erase the underlying medical need. While waiting on appeal, patients must either pay out of pocket or forgo care, potentially becoming sicker and facing higher costs later. These added costs ultimately ripple through the health care system, contributing to higher prices and premiums. Limited transparency makes it difficult for patients and providers to understand or challenge these decisions, increasing the risk that wrongful denials will delay care and worsen health outcomes. Patients with chronic illnesses may be especially vulnerable to these denials. Black and Hispanic patients, other people of color, and those who identify as lesbian, gay, bisexual, or transgender face higher rates of claims denials.
AI can also increase health care spending through the tools providers use to document and bill for care. AI-powered medical scribes and coding tools are designed to improve clinical documentation and efficiency, but may also contribute to “upcoding” – billing for more complex or expensive services than were actually provided. PwC lists AI-enabled documentation and coding tools first among the biggest drivers of health care costs in 2027. AI systems may document additional symptoms that support more severe diagnoses, default to higher billing levels, or classify routine visits as complex encounters based on patient conversations. Evidence suggests these practices can have significant financial consequences. A Blue Cross Blue Shield Association analysis estimated that roughly $2.3 billion in spending nationwide may be tied to AI-powered coding. Coding of a single condition in maternity admissions alone added $22 million in one year at the hospitals studied.
Taken together, these trends illustrate a broader paradox. AI is often promoted as a tool to reduce administrative waste and improve efficiency. But when deployed primarily to optimize insurer profitability or maximize reimbursement, the technology can instead increase administrative complexity, amplify inequities, and drive higher health care spending.
Across nearly every dimension of life, companies use AI systems to collect and analyze personal and behavioral data to determine the price people pay. Whether it is called surveillance pricing, dynamic pricing, or algorithmic pricing, the underlying reality is the same: companies are deploying AI systems to analyze consumer data and tailor prices to the individual. Instead of one market price, firms are experimenting with millions of personalized prices designed to extract the maximum possible amount from each person. An investigation of Instacart pricing by Groundwork Collaborative, Consumer Reports, and More Perfect Union found nearly three-quarters of grocery items were offered to shoppers at multiple price points, prices were up to 23% higher for some shoppers on the exact same items, and these practices could translate into a swing of about $1,200 a year for a family of four.
Data brokers sit at the center of this system. They collect, analyze, and sell personal information on millions of internet users, drawn from targeted advertising, social media platforms, and online activity. The data is then packaged and sold to corporations that use it to shape marketing strategies, predict consumer behavior, and increasingly to determine individualized prices. AI-driven pricing systems are spreading rapidly across the economy. This report focuses on some of the most immediate impacts on household costs, including car insurance, airline tickets, and hotels.
Although a handful of states have begun to address surveillance-based pricing practices, their efforts remain fragmented and insufficient to confront the problem at scale. Maryland and New Jersey recently enacted laws to curtail personalized grocery pricing, and while Connecticut passed broader legislation, it doesn’t take effect until July 2027 and may be amended before then. Lawmakers in Colorado passed a more expansive ban, though Gov. Jared Polis vetoed the measure in June. A similar bill is awaiting signature on New York Gov. Kathy Hochul’s desk. In total, roughly 90 bills in 27 states to curb algorithmic pricing have been introduced since 2025, more than 40 of them this year. However, a federal solution is needed to protect all consumers.
AI is changing car insurance by allowing companies to set premiums using increasingly detailed data about how, when, and where people drive. Modern vehicles and driving apps generate enormous amounts of behavioral data that are shared with automakers, data brokers, and insurers. This information – known as telematics – includes speed, braking patterns, acceleration, and location tracking. Insurers analyze this information with AI to assess risk and set personalized premiums. While this may reward some safe drivers, it can also lead to higher premiums and be used to deny coverage entirely.
Telematics-based pricing can lower costs when drivers knowingly agree to share their data. Through usage-based insurance programs, drivers opt in to monitoring in exchange for premium discounts that averaged about 12% in one large insurer’s program. But that is different from the growing practice of collecting driving data without clear consent and selling it through intermediaries. Companies such as LexisNexis Risk Solutions and Verisk aggregate driving behavior data and sell it to insurers, often without drivers realizing their vehicles are generating a detailed behavioral record. By its own account, LexisNexis products were used in 86% of new U.S. auto insurance policies issued in 2023.
These systems are far from perfect. AI models can misinterpret data, misclassify normal driving behavior as risky, or rely on incomplete datasets. When that happens, drivers may face higher premiums based on flawed or misleading information. And while AI may reduce insurers’ costs by speeding up claims processing, there is little evidence that those savings are passed on to consumers.
The consequences are already obvious. Car insurance premiums have risen about 50% since early 2022, with the average annual cost of full coverage above $2,200. While rising vehicle costs, expensive replacement parts, and higher medical expenses after accidents contribute to these increases, surveillance-driven pricing is adding another layer of financial pressure. More than half of drivers now say car insurance is a “financial burden.”
Premiums have eased slightly since peaking in early 2026, after several years of steep growth. But even if prices stabilize, AI-driven pricing systems will continue reshaping how premiums are determined. Personalized pricing may reduce costs for some drivers while sharply increasing them for others.
Between airfare and lodging, families face steep travel costs that can easily eat up a significant chunk of a household’s budget. AI-driven pricing systems are making travel even more expensive by allowing travel companies to collect personal information, track online behavior, and aggregate data across platforms to tailor prices to individual consumers based on their maximum willingness to pay.
Airlines are increasingly adopting these systems. By analyzing demand, search activity, booking patterns, and other consumer data, airlines can continuously adjust fares rather than offer one transparent market price. Delta CEO Ed Bastian said in August that AI could raise the airline’s profitability by as much as 50%, lifting margins from about 10% to 15%. Delta partners with AI-powered pricing tech firm Fetcherr, which claims its “secret sauce” is all the data they can get their hands on.
Households spent an average of $682 on airline fares in 2024, and these pricing practices can add to a significant annual expense. Even if families aren’t regularly traveling for leisure, these systems can jack up prices when they need to fly for emergencies. In one widely shared case, a customer booking a flight to a funeral saw a JetBlue ticket price jump $230 in a single day. In response, JetBlue’s official social media account recommended that the customer use incognito mode to book the flight, suggesting that the price may have been set using the customer’s web browser data.
Hotels are increasingly using the same surveillance pricing techniques. AI systems adjust room prices in real time based on demand signals such as conferences or major events – and increasingly, on who is looking. AI-driven personalization and upselling can increase room rates in hidden and explicit ways. In a December 2024 test, popular booking sites showed higher hotel prices to travelers browsing in the San Francisco Bay Area than to travelers checking rates in cities like Phoenix or Kansas City. In 2012, Orbitz discovered that Mac users spent as much as 30% more a night on hotels and responded by steering them toward pricier options than those displayed to other users.
Hotels are also expanding surveillance inside the room itself. Some hotels now use monitoring systems that track noise levels and vape particles to impose additional fees on guests.
These practices make it increasingly difficult for travelers to know whether they are seeing a market price or a price tailored to extract more from them. As AI-driven pricing spreads across airlines, hotels, and travel platforms, families may pay more for the same trip without knowing how or why their price was determined.
The rapid expansion of AI-driven data centers is placing enormous strain on the U.S. electricity system, and ordinary households are paying the price. There are now more than 5,000 data centers in the U.S., and their electricity consumption is growing at a staggering pace. In 2023, data centers consumed roughly 4.4% of all electricity in the country. By 2030, that share could nearly triple to about 12%. This surge is reshaping energy markets, utility planning, and electricity prices across the country.
Data centers raise energy costs in two fundamental ways. First, they consume massive amounts of electricity locally, forcing utilities to expand transmission lines, substations, and generation capacity. Those infrastructure costs are often spread across the entire customer base, meaning households and small businesses subsidize the build-out required for these facilities. Second, the sheer scale of their electricity consumption increases overall demand in regional markets, pushing wholesale prices upward and driving up utility bills for everyone else.
The consequences are clear in household energy bills. Electricity prices have risen more than 40% since 2019, driven in significant part by the expansion of AI data centers. In areas with heavy data center activity, wholesale electricity prices in some months have increased by more than 250% in the past five years. Residential electricity rates could climb 15% to 40% above 2025 levels by 2030, depending on the utility.
Much of this growth is driven by “hyperscale” data centers, facilities that use more than 100 megawatts of electricity. For perspective, 100 megawatts of power is enough to supply roughly 80,000 U.S. households. There are already hundreds of these hyperscale facilities operating across the country, and roughly half of the 1,500 large data centers in the global pipeline are planned for the United States. These large-scale centers are heavily concentrated in a handful of regions, including Northern Virginia, Phoenix, Dallas, Atlanta, and Chicago, meaning the burden of their electricity demand is falling disproportionately on the residents of those areas.
Northern Virginia illustrates the scale of the problem. The region is the world’s hyperscale data centers, and data centers there consume over a quarter of the state’s total electricity supply. In April 2025, the state’s largest utility, Dominion Energy, proposed its first base-rate increase since 1992. In November 2025, regulators ultimately approved rate hikes that add about $11.24 per month in 2026 and another $2.36 in 2027. Similar pressures are emerging across the country. In the Midwest and Mid-Atlantic region, analysts project that data center-driven electricity demand could add roughly $18 a month to residential bills in western Maryland and $16 in Ohio – roughly $190 to $220 a year. Residents are footing the bill for Big Tech’s buildout.
The costs are not limited to higher utility bills. Data centers also impose significant public health and environmental consequences. Their electricity demand increases emissions from power plants and backup generators, worsening local air pollution and increasing respiratory illnesses in surrounding communities. Nationwide, the health damages associated with data center-related pollution could reach $20 billion a year by 2028, along with more than 1,000 premature deaths a year. Backup generators at Northern Virginia data centers alone could impose $190 million to $260 million a year in public health costs across Virginia and neighboring states and contribute to approximately 14,000 asthma flare-ups and breathing episodes.
The AI buildout is projected to cost $7.6 trillion across chips, data centers, and power between 2026 and 2031. Big Tech has the capital to build the infrastructure it needs many times over even as households are being asked to help finance it through rate hikes. A Groundwork proposal to freeze and stabilize household electricity rates nationwide – alongside public power investments and data center accountability measures – would cost taxpayers a modest $137 billion over 10 years and save households an estimated $255 billion over that same span. The price tag for protecting families from data center-driven rate hikes is a small fraction of what the AI industry alone stands to earn.
Despite these rising costs, state and local governments continue to provide massive tax subsidies to data center developers. Thirty-seven states now offer tax incentives for new data centers, often lasting 10 to 50 years, and the scale of these subsidies is enormous. In Indiana, a deal with Amazon Data Services is valued at more than $8 billion, equivalent to nearly $1,200 for every resident of the state. Another deal between Amazon.com and Morrow County, Oregon, is worth roughly $1 billion, or nearly $80,000 per person in the county. In Texas, a 2023 incentive package for Meta Platforms Inc. in El Paso is estimated at nearly $700 million, or about $1,000 per resident of the city.
In short, communities are being asked to subsidize infrastructure, absorb higher electricity bills, and bear the public health costs of pollution to support the data center industry. Data centers may be powering the AI economy, but the public is bearing the costs.
The AI boom is beginning to make everyday technology more expensive. Historically, technological advances allowed manufacturers to offer better performance without comparable price increases, making consumer hardware a largely deflationary part of the economy. That pattern began to reverse in the first half of 2026, as surging demand from AI data centers pushed up the price of memory and other components also used in consumer devices.
AI data centers require enormous quantities of memory, storage, and advanced chips to train and operate increasingly powerful models. Many of these components are also essential to computers, phones, gaming consoles, and other consumer electronics. As technology companies expand their AI infrastructure, consumer-device manufacturers must compete for the same limited supply.
Higher component prices then flow through the supply chain and raise the cost of producing everyday electronics. Major technology companies have already begun passing these costs on to consumers. Since the start of 2026, Apple has raised prices on some of its most popular laptops by $200 to $300 and on every iPhone by at least $100, and Microsoft has increased Xbox console prices by $100 to $150. This makes it harder for families to afford the routine purchases they can’t easily put off, like replacing a laptop needed for work or school, or basic home appliances that rely on the same cheap memory chips now being diverted to AI data centers.
AI is transforming the economy at extraordinary speed. While the technology holds enormous potential, its early deployment is already reshaping markets in ways that often increase costs for households and accelerate the transfer of wealth from everyday Americans to AI executives and their shareholders.
AI is enabling more sophisticated fraud schemes, amplifying the mental health harms associated with social media, reshaping insurance practices, spreading personalized pricing across industries, and increasing electricity demand through massive data centers. In many cases, these costs are hidden inside complex systems of algorithms, data brokers, and automated decision-making, leaving consumers paying artificially inflated prices for everyday goods and services.
Despite the growing economic footprint of AI, reliable public data on its real-world impacts remain limited. Much of the information needed to understand how AI affects prices, wages, and household finances is controlled by the companies developing and deploying these systems.
To craft effective policy, policymakers need to demand better data and stronger transparency requirements. Several proposals from research organizations such as SeedAI and the Brookings Institution have called for the creation of a Generative AI Intensity Index to track how AI systems are being used across the economy. At the same time, policymakers must move quickly to ban the practice of surveillance pricing, prohibit insurers from using AI to deny health care, prevent data center developers from driving up household energy bills, and adopt stronger liability frameworks that can hold AI companies accountable when their products cause financial or physical harm.
As AI reshapes the economy, the deeper question is one of democratic control over this new technology. Right now, a small number of companies decide how AI is built and where it is deployed, and they answer to shareholders – not to the households absorbing the costs. Elected officials have the authority to change that. They can set enforceable rules for how these systems are used, guarantee human review of automated decisions about people’s health, work, and finances, and require independent scrutiny of the most powerful models before they reach the public. AI’s trajectory should be set by the public through its representatives, not by the companies profiting from it.