The 2026 U.S. midterm campaigns are the first in which AI-generated political ads are regularly appearing on people’s televisions and social media feeds.

We are researchers who have been studying political advertising through the Wesleyan Media Project since 2010. This election cycle – using data from media reports, student coders and AdImpact, a firm that tracks political ad spending – we’ve tracked about US$80 million in spending on almost 170 unique ads that use AI.

What we found has surprised us: AI use spans from hyperrealistic deepfakes to subtle enhancements; Republican sponsors – both candidates and interest groups – are much more likely to use AI than are Democratic sponsors; and, thanks to a patchwork of state legislation, many of these ads do not disclose the use of AI at all.

From deepfakes to subtle edits

An assortment of politicians and watchdog groups have expressed concern about campaigns using generative AI to produce deepfakes – synthetic videos showing people doing things they did not do – that might deceive voters.

We’ve noticed several ads containing hyperrealistic deepfakes of famous politicians, including Donald Trump, Nancy Pelosi, Barack Obama, Kamala Harris and Alexandria Ocasio-Cortez. Ocasio-Cortez, in particular, is a favorite among Republican advertisers, appearing in at least five ads.

We’ve seen deepfakes in which a Republican Senate candidate from Louisiana drives a school bus full of undocumented immigrants, a Republican candidate for governor from South Carolina walks arm in arm with drag queens, and an ad in which Liz Cheney, Mitt Romney and Mike Pence are seen carrying pitchforks on the White House lawn.

People who are not politicians made appearances, too, including a fake Dr. Anthony Fauci, seen running around a state fair with a huge syringe, presumably eager to vaccinate everyone. We’ve also noticed several ads in which AI was used to generate crowds or constituents.

In several cases, AI was used to enhance visuals rather than generate something new. One ad from Chip Keating, a Republican candidate for governor in Oklahoma, includes an AI disclaimer, but it doesn’t specify exactly how AI was used. Ads that use AI to enhance visuals don’t necessarily look different from ads that were created in the pre-AI era, which makes it difficult for viewers to discern whether they depict something false.

A partisan gap

Republicans – both candidates and groups such as super PACs and 501(c) organizations – are much more likely to use AI in their ads than are Democrats, according to our research.

In fact, Republican candidates or pro-Republican groups were behind 80% of the ads we tracked and 83% of the spending.

---

Start your day with essential news from Salon.
Sign up for our free morning newsletter, Crash Course.

---

We can only speculate as to why Republicans dominate the use of AI in political advertising in 2026. In general, Democrats tend to take on a regulatory mindset when it comes to political campaigns, favoring limits on campaign spending and required disclosures. In 2022, for instance, only Senate Democrats and two independents voted to advance the DISCLOSE Act that would have required additional campaign finance disclosures for super PACs, labor unions and corporations. Republicans, by contrast, tend to be more in favor of a free market approach.

These more general philosophies may be reflected in the parties’ use of generative AI for political advertising, something about which voters are worried. Polling shows broad support for more regulation, with 78% of registered voters favoring a ban on AI content that makes deceptive claims about candidates.

        

Disclaimers all over the map

Because regulation of AI in advertising depends on a patchwork of state legislation, many of these ads are not required to disclose the use of AI. This lack of disclaimers makes tracking AI use in ads challenging. Our team has relied on media coverage and trained student coders to flag ads that are potentially AI-generated.

Across 35 states, only 31% of the ads we tracked – representing 22% of the spending – disclosed the use of AI tools. The wording of these disclaimers was all over the map. For example, one Georgia ad included the disclaimer, “This video has been manipulated or generated with artificial intelligence,” while an Oklahoma ad said, “Political satire. AI-generated images do not depict actual events.” A North Carolina state Senate ad said, “You guessed it! AI was definitely used to generate these silly video clips.”

Laws don’t drive disclosure

Some advertisers voluntarily disclose the use of AI even when they are not required to do so. Sometimes, the opposite occurs – advertisers don’t include disclaimers even when state law requires them to. In fact, we’ve found that a state law that requires disclaimers on ads that use AI has very little relationship with the actual use of disclaimers.

In states without laws governing the use of AI in political ads, 32% of ads contained a disclaimer; by contrast, in states with laws governing the use of AI, 29% contained a disclaimer. This comparison, however, is not perfect, as some of the ads in our database aren’t covered by their state’s law. Some states, such as Colorado, have laws that apply only to candidate deepfakes and thus exclude other uses of AI. In other instances, such as in Louisiana, an AI law was enacted after the ad aired.

Minnesota law generally bans deepfakes in political ads, but an ad featuring synthetic video of Democratic Senate candidate Peggy Flanagan aired in May 2026 anyway. Whether it violated state law is uncertain, as the law requires that the media be “so realistic that a reasonable person would believe it depicts speech or conduct of an individual who did not in fact engage in such speech or conduct.”

Moving forward, the real policy challenge will revolve around transparency, the enforceability of existing laws and simply figuring out what type of disclosure would be helpful to voters.

Travis N. Ridout, Professor of Government and Public Policy, Washington State University; Erika Franklin Fowler, Professor of Government, Wesleyan University, and Michael Franz, Professor of Government, Bowdoin College

This article is republished from The Conversation under a Creative Commons license. Read the original article.

As Americans prepared to celebrate the nation’s 250th birthday on the morning of July 4, sobering images flashed across their phones and TV screens. Hundreds of masked men marched the streets of Washington carrying flags and chanting “Reclaim America!” It soon became clear from the logos on their identical baseball caps that the marchers were members of Patriot Front, a white nationalist group.

The march ended without incident. Still, it joined the 2017 “Unite the Right” rally in Charlottesville, Virginia, and the Jan. 6, 2021, attack on the U.S. Capitol – both of which culminated in deadly violence – as recent mass demonstrations with participants who openly embrace white nationalism. Members of a loosely organized movement that usually operates out of public view, white nationalists hold that shifting demographics pose an existential threat to the white race, which they view as culturally superior and deserving of power over other groups in the U.S.

Combined with the spread of white nationalist ideology from obscure corners of the internet into mainstream politics, these events have led many Americans to wonder: How extensive is white nationalism’s appeal among the broader public, and who is drawn to this extremist ideology?

We are political scientists who study political behavior, identity and social media. In recently published research in the scientific journal Nature, we offer some answers to these questions using national surveys we conducted with nearly 8,000 non-Hispanic white Americans from 2021 to 2025.      

Asking Americans about white nationalism

We expected many Americans would be unfamiliar with the term “white nationalism,” so we defined it for survey respondents with a brief statement covering the movement’s major tenets: the purported need to maintain a white majority in the U.S., claims about the superiority of white culture, and calls for white people to hold more political and economic power than other groups. We measured endorsement of the movement’s ideology, not participation in the movement itself or approval of its tactics.

We randomly varied whether respondents saw the description before or after we asked whether they supported the white nationalist movement. Overall, nearly 5% of white adults – about 1 in 20 – said they did.

Those asked the question before seeing the description were less likely to express support at first, but their reported support rose after they read it. Along with several tests for possible response bias, this suggests that 5% is a reasonable lower-bound estimate of the true level of support.

Whether 1 in 20 white adults is a high or low number is a matter of interpretation. For context, it’s roughly the same as the share of American adults who identify as vegetarian or vegan.

Young white men

A number that jumped out to us was the share of white men under age 30 who support white nationalism: just over 13%, or about 1 in 7. This was more than twice the level of support for white nationalism among women in this age group: 6%. Endorsement of white nationalism among men significantly declined by age: Only 2% of white men age 65 and over registered support.

---

Start your day with essential news from Salon.
Sign up for our free morning newsletter, Crash Course.

---

Our discovery that white men born before the passage of the Civil Rights Act were more likely to reject white nationalism than those coming of age today may come as a surprise, as younger white Americans typically hold more tolerant views on racial matters than their elders. But young men have long been overrepresented in extremist groups, and white nationalism’s grievance narrative may resonate for young white men today as they fall behind previous generations of men on key indicators like finishing college, starting careers and forming families, as well as on their overall health. 

Online life

Another key predictor of support: online social life that crowds out offline relationships. We asked respondents to call to mind their three best friends and then to tell us whether each friend was known only offline, only online or both.

We found that the more best friends known only online, the stronger one’s support for white nationalism. The internet and particularly social media are channels through which people may be exposed to white nationalist content. They are also places where alienated people may turn for the human connections they lack offline.

Interestingly, we did not find a general relationship between more social media use and more support for white nationalism. So while online life appears to be related to support for white nationalism, what matters in our findings is not how much people use social media, but whether their closest friends are known only online.

Individual and social strain

Strain, whether defined as undergoing personal difficulties or being exposed to wider social distress, is also a strong predictor of support for white nationalism. We asked respondents whether they had suffered from specific hardships over the past year, including the death of a loved one, losing a job, being the victim of a crime or getting divorced. Respondents experiencing more of these adversities were more likely to endorse white nationalism.

Social strain in the counties where respondents live – such as poverty, unemployment and population loss – emerged as another strong predictor of support for white nationalism, as did countywide rates of “deaths of despair,” which include suicide and deaths associated with drug and alcohol use. In other words, support for white nationalism was higher among people living in more distressed or economically depressed places.

More than just partisan politics

White nationalism is often cast as a political phenomenon by advocates, commentators and scholars, and indeed it is.

We found in our research that white Democrats and liberals are significantly less likely to endorse white nationalism than white Republicans and conservatives. However, we also found high levels of support among those who were unsure of their ideology or partisanship, suggesting a potential untapped reservoir of political support for white nationalist ideas.

Still, our findings show that support for white nationalism cannot be explained solely by partisan politics. Larger forces, including generational change, the ubiquity of the internet and the strain Americans are experiencing in part due to profound economic transformation are at work.

In our view, those who seek to understand and address this critical development in American public life must reckon with the broad structural shifts that have reshaped Americans’ identities, social relations and sense of economic security and that undergird white nationalism’s appeal to the broader public.

Patrick Egan, Professor of Politics & Public Policy, New York University; James Bisbee, Assistant Professor of Political Science, Vanderbilt University, and Joshua A. Tucker, Professor of Politics, New York University

This article is republished from The Conversation under a Creative Commons license. Read the original article.

For defendants hoping to avoid jail and walk away from millions of dollars in financial sanctions, paying, say, US$600,000 to a lobbying firm to influence the president may seem like a good investment.

House Democrats recently reported that Trump’s second-term clemency – which includes both pardons, which wipe away every part of a conviction, and commutations, which typically leave convictions intact but shorten a person’s sentence – has saved convicted defendants nearly $1.7 billion in forgiven financial obligations.

Some of that amount – about $700 million – had not yet been officially imposed by courts because the clemency came through so quickly. Much of the remaining $1 billion had been ordered as restitution, meant to repay victims’ losses.

As a legal scholar researching the evolution of modern presidential clemency, I have reviewed every individual pardon and commutation granted by U.S. presidents since 1989, spanning seven administrations, from George H.W. Bush to Trump’s second term. I have found that while other presidents have canceled tens of millions of dollars’ worth of restitution and fines, none of them have come anywhere near the billion-dollar mark – until now.

What is restitution?

The main financial penalties faced by federal defendants are fines, forfeiture – where a defendant has to give up, say, the Ferrari they bought with the profits from their drug business – and restitution.

Restitution is an order to pay back the criminal’s gains to the victims of the crime. For example, if the CEO of a company tricks people into investing $600 million into a hydrogen-powered truck that was never operational, the court may order that, if convicted, he pay back the people he defrauded.

Restitution is rarely able to return all of someone’s lost money, but it is supposed to prevent the fraudster from enjoying the fruits of their crime.   

Getting away with it

Since January 2025, Trump has granted clemency to a lengthy list of big-ticket fraudsters, including a cryptocurrency billionaire whose exchange pleaded guilty to anti–money laundering violations and CEOs convicted of payroll tax fraud.

---

Start your day with essential news from Salon.
Sign up for our free morning newsletter, Crash Course.

---

Many of these pardons and commutations have coincided with the use of lobbying firms or generous donations to causes close to Trump’s heart: his inaugural committee, his reelection campaign or MAGA Inc. One successful pardon-seeker’s mother attended a $1 million fundraising dinner.

Clemency seekers use a variety of ways to catch the president’s attention. The end result is that victims of crime – and taxpayers, who have to pay for these huge white-collar investigations and trials – are often left with nothing.

Unfettered power

The U.S. Constitution grants the president nearly unbounded power to pardon whomever he likes, whenever he likes and for whatever reason he likes. This is why, since 1894, the Office of the Pardon Attorney has tried to impose some guidelines on the process and decide which cases should be eligible for clemency at all.

These guidelines generally require that a clemency seeker have been released from prison for five years before they are eligible to request a pardon. If they never went to prison, the five years is calculated from the date of sentencing. Many presidents adhere to this norm: My study shows that between 94% and 100% of people pardoned by George H.W. Bush, George W. Bush and Barack Obama met this criteria. Because he gave so many preemptive pardons, Joe Biden’s percentage was lower, at around 68%, but still well over half.

In contrast, out of the 216 individual pardons granted by Trump in his second term so far, only 34 – about 15% – have gone to people who waited that long.

The money ladder

A look back at recent past administrations shows an erratic but increasing amount of financial obligations canceled. While some presidents have been frugal with their financial forgiveness – George H.W. Bush and George W. Bush remained under the million-dollar mark, while Joe Biden was just over it – others have been less so.

Bill Clinton erased nearly $10 million in fines and restitution through pardons, most notoriously pardoning Marc Rich, a fugitive financier whose ex-wife contributed more than a million dollars to various Democratic causes and the Clinton Presidential Library.

Obama more than doubled Clinton’s record, forgiving $23 million in financial sanctions.

But Trump surpassed them both in his first term, canceling nearly $195 million in financial obligations. The amount of money saved by convicted defendants during his second term is exponentially higher. One could even say that, as Trump likes to put it, “We’ve never seen anything like it.”

A new role for commutations

Since the main function of commutations is to let people out of prison sooner, commutations usually leave financial obligations in place. During his first term, Trump primarily granted commutations that left “intact and in effect” all conditions of a person’s sentence except for imprisonment. This let numerous defendants out of jail but kept them on the hook for nearly $3 billion in fines and restitution.

That has changed during Trump’s second term. Today, his commutations routinely cancel a recipient’s prison sentence “with no further fines, restitution, probation or other conditions.”

This is how, through commutations alone, Trump has canceled more than $300 million in penalties and restitution since he returned to office.

A word of caution

One figure that is unavailable in public records is whether any given defendant has started to pay, or has fully paid off, their financial obligations. So these figures for every administration represent a ceiling: the full amount of fines, restitution or forfeiture that a judge imposed on a convicted defendant. How much was actually paid down can usually only be estimated, though there are exceptions.

For example, Jason Galanis, who pled guilty to defrauding the Oglala Sioux Nation, had paid just over $2 million of his $80 million restitution before his sentence was commuted and the restitution erased in 2025. But we only know that because he unsuccessfully asked the court to return the money.

One useful indicator of how much someone may have already paid is the speed at which they got clemency. The faster they receive clemency, the less time they have to pay off their penalties. The median delay between sentencing and receiving a pardon during Biden, Obama and George H.W. Bush’s administrations was over 20 years.

During Trump’s second term, in contrast, things are moving faster. Trump has pardoned or given commutations to at least 44 recipients within a year of their sentences who together account for over $270 million in financial obligations.

This includes a nursing home executive who pled guilty to withholding his employees’ payroll taxes – which prosecutors said he spent on a yacht – and was pardoned 12 days after being sentenced, ensuring that he could avoid some $4 million in restitution. A crypto company was pardoned hours before it was due to pay a $100 million fine for violating anti–money laundering laws. A digital media founder was ordered to repay defrauded investors $36 million. His sentence was commuted, and his restitution canceled within four months of sentencing.

In these cases, clemency was issued before most of the money was able to reach the government or the victims it was owed to. The faster clemency is granted, the more likely it is that fraudsters get to walk away without paying their victims anything.

Caren Morrison, Professor of Law, Georgia State University

This article is republished from The Conversation under a Creative Commons license. Read the original article.

On Sept. 19, 2026, journalists from CNN, MS NOW, and Politico were denied access to the White House after President Donald Trump revoked their credentials, citing “their constant ‘reporting’ FAKE NEWS!” On social media, Trump posted, “Media Outlets shouldn’t be able to constantly write or report FICTION and LIES when they’re covering the President.”

This is not Trump’s first attempt at banning journalists from the White House. During his first term, he sought to ban CNN’s Jim Acosta by revoking his credentials following a contentious news conference about the 2018 midterm elections.

In that conflict, CNN quickly filed a lawsuit, arguing that, “If left unchallenged, the actions of the White House would create a dangerous chilling effect for any journalist who covers our elected officials.”

Journalism organizations also responded. The Committee to Protect Journalists, the Radio Television Digital News Association and even Acosta’s peers at The Daily Caller – a right-leaning news outlet – condemned the White House ban.

A federal court granted CNN a temporary injunction – reinstating Acosta’s press pass. The Trump administration later restored Acosta’s credentials, and CNN dropped its lawsuit.

Many commentators and members of the press have called Trump’s latest ban on the three news organizations a violation of the First Amendment.

White House Correspondents’ Association President Jacqui Heinrich said on X, “Today’s action revoking access for journalists from CNN, POLITICO, and MSNOW violates the First Amendment.”

Even George Washington University law professor Jonathan Turley, who staunchly supports Republicans and frequently criticizes the press, said on X that the ban “would be an assault on long-honored principles of a free press.”

As Turley and many others have argued over the weekend, Trump’s ban is almost certainly unconstitutional under the First Amendment, which states that “Congress shall make no law … abridging the freedom of speech, or of the press.” Media law professor Frank LoMonte argued in 2018, “The First Amendment guarantees all Americans freedom to speak and publish without government interference.”

But they may be missing a crucial point. In fact, there’s more than one constitutional right involved in such a case. The judge who restored Acosta’s press pass in 2018 primarily based his ruling on the reporter’s Fifth Amendment rights to due process, not on the First Amendment.

---

Start your day with essential news from Salon.
Sign up for our free morning newsletter, Crash Course.

---

And in my view as a constitutional law professor, it is less clear whether the press has a constitutional right of access to the White House even though past presidents have honored and respected the press’s access to the executive branch.

While several Supreme Court cases, such as Richmond Newspapers v. Virginia, unequivocally provide the press with a right of access to criminal trials, there is no Supreme Court precedent that specifically provides a clear constitutional right of access to the White House.

Make that two constitutional rights

In one case from 1977, Sherrill v. Knight, Robert Sherrill, the Washington correspondent for The Nation, was – after holding press passes over several years – denied a press pass because the Secret Service advised he should not be given one. The Secret Service refused to disclose their justification.

Sherrill sued, and the D.C. Circuit Court of Appeals concluded that the Secret Service cannot deny a White House press pass absent “narrow and specific standards by which applications are to be judged.”

The court required the White House to provide the press with “notice of the factual bases for denial, an opportunity for (them) to respond to these, and a final written statement of the reasons for denial (of a press pass).”

In the court’s view, these requirements – notice and an opportunity to rebut the bases for the denial of a press pass – are protected by two constitutional guarantees working together, the press protections in the First Amendment and the Fifth Amendment’s protections from arbitrary deprivations of “life, liberty, or property, without due process of law.”

In other words, before the federal government can deprive an individual of their life, liberty or property, the federal government must provide that individual with due process, whose purpose is widely accepted as preventing “arbitrary governmental actions.”

Due process is key

There are a number of Supreme Court decisions that could be key in the current journalists’ lawsuit, which was filed on Sept. 21, 2026.

Among them is Mathews v. Eldridge from 1976, in which the Supreme Court stated that due process is flexible depending on the extent of the liberty or property interest at stake. But in general, the justices said, due process requires the government to provide notice to the individual and an opportunity for the individual to present their case before the deprivation occurs.

In Sherrill v. Knight from 1977, the court ruled that the denial of a press pass implicated a First Amendment liberty interest. That interest, the court wrote, “undoubtedly qualifies as liberty which may not be denied without due process of law under the fifth amendment.”

This nuance – the First and Fifth Amendment working in tandem – is what most commentators and critics are missing in their response to the recent ban of CNN, MS NOW and Politico. In fact, these outlets collectively lead with their claim that the Trump administration violated their Fifth Amendment right to due process in their lawsuit to block Trump’s ban.

“Even if attending a White House event is not a constitutionally guaranteed right, that doesn’t mean the government is free to take it away as punishment,” LoMonte argued in 2018.

Based on public reporting, it is clear that the Trump administration did not provide CNN, MS NOW or Politico with clear written notice stating the reasons for the revocation of their press passes. It certainly did not provide the outlets with an opportunity to rebut Trump’s unsupported claim that they report “FICTION and LIES.”

The same Trump-appointed judge who decided the CNN Acosta case has been assigned to the latest press ban case. So, if the CNN Acosta case is any indication, I expect the federal judge to promptly, but temporarily, reinstate the news outlets’ press passes while Trump tries to argue that his actions are constitutional, which in my view, they are not.

Wayne Unger, Associate Professor of Law, Quinnipiac University

This article is republished from The Conversation under a Creative Commons license. Read the original article.