This week, as America marks its 250th birthday, political violence nationwide is at its sharpest sustained escalation in decades. PERIL’s Director of Research tracked a roughly 40% rise in terrorism and targeted-violence incidents in the first half of 2025 alone compared with the same period a year earlier, and the Bridging Divides Initiative says the risk environment has continued to worsen into 2026.

Millions of Americans use AI chatbots daily, and some of those conversations involve violence — as fantasy, as research into methods, or as a real step toward an attack.

In March, the Center for Countering Digital Hate (CCDH) and CNN’s Investigations Unit posed as people signaling intent toward violence — school shootings, knife attacks, political assassinations, bombing attacks on synagogues and political parties — across 10 mainstream chatbot platforms. Eight of the 10 platforms helped in more than half of the test cases, going so far as to provide advice to the users on locations they could target and weapons they could use. Only two of the platforms did not.

The risk is not theoretical. Investigators looking into the January 2025 Las Vegas Cybertruck explosion found that the perpetrator had used a chatbot to research explosives. A Finnish teenager spent months in 2025 using a chatbot to refine a manifesto, which included his detailed attack plan, before carrying out a school stabbing. And the families of children killed and wounded in a February 2026 mass shooting in Canada are now suing an AI company, alleging its chatbot did not challenge the shooter or direct her to seek help, as well as that the company failed to report the chats to authorities.

Meanwhile, researchers at Northeastern University found that the refusal and de-escalation behaviors chatbots are trained to use often deactivate the moment a user slightly reframes their intent — even after stating outright that they want to hurt themselves or someone else. And in a 2026 APA survey, nearly nine in 10 psychologists said they worry chatbots may inadvertently encourage self-harm — missing genuine crisis signals in the people using them.

So, the bottom line is that millions of people, including some on an identifiable pathway to violence, are having exactly the kind of conversation that a trained human would recognize as a warning sign — and having it with a system that, by design or by neglect, frequently fails to notice or intervene.

We already know how to fix this problem.

This isn’t a call to make chatbots into therapists or threat assessors. It’s a call to apply something we already know works: the friction-and-referral model.

Recall from our last issue of The Brief that X’s own testing found that a simple “are you sure?” prompt caused over a third of people to revise or cancel a hostile reply, with lasting effects on their behavior afterward. Introducing these prompts — often referred to as cognitive interrupt design — does not equal a moral judgment embedded in the product. It’s a pause, offered at the exact moment judgment is most likely to be compromised, paired with a door to a real off-ramp, such as a crisis line or a link to an organization like Life After Hate.

Some labs are already doing versions of this for suicide risk specifically. OpenAI, for example, has said publicly it has worked with mental health clinicians to help its models recognize distress and guide people toward professional care. While we’re certainly far from where we should be, it’s a start. And thankfully, states are moving too:

• New York and California currently require detection and referral protocols for suicidal ideation and self-harm specifically as well as require chatbot operators to refer users to crisis service providers when a user demonstrates certain indicators.

• Texas’s new Responsible AI Governance Act explicitly bars AI systems from intentionally encouraging “self-harm, violence, or criminality” — violence named outright, not folded into self-harm.

• Connecticut’s newly passed SB 5 requires operators of AI companions to implement evidence-based protocols to detect users’ expressions of a risk of suicide, self-harm, or imminent physical violence. Those operators must also prevent the AI companions from generating outputs that encourage such activity.

• Pennsylvania’s proposed SAFECHAT Act would require companies building AI companions to similarly implement a protocol, described on the companies’ websites, to prevent the companion from generating suicidal ideation, suicide, or self-harm content as well as content that encourages people to commit violence. The companies must have the companion provide a notification that refers the user to crisis service providers should the users express suicidal ideation, suicide, or self-harm, though not if they express an intent to commit violence.

• Law firms advising chatbot operators are already telling clients to calibrate detection for intimations of suicide, self-harm, or violence to others.

• The EU’s AI Act requires that the developers of general-purpose AI systems that pose what the EU describes as “systemic risk” conduct adversarial tests of models — and document those tests. A separate requirement — that chatbots notify users they’re talking to AI, not a human — takes effect this August.

Clearly people are thinking about this. But we still have a lot of work to do.

One critical element of what’s still missing is the connective tissue to the research base that makes any of this precise rather than performative; for example, the pathway-to-violence indicators that the Secret Service’s National Threat Assessment Center has spent nearly three decades documenting. Such indicators include leakage of intent, grievance fixation, and escalating research into means. However, the NTAC itself (which is part of DHS) has not yet issued public guidance connecting its own threat-assessment framework to AI chatbots, even as its research is among the clearest and most authoritative existing blueprints for what “concerning indicators” should actually mean in this context.

This isn’t new territory for AI. The tech stack exists.

The infrastructure and the precedent both already exist. AI companies already route people to trusted outside resources contextually in several cases:

• Elections. OpenAI will partner with Democracy Works for the 2026 U.S. midterm elections to provide ChatGPT users with authoritative voting information — including polling locations, registration and voting deadlines, ballot information, plus live results from the Associated Press — whenever someone asks an election-logistics question. Anthropic has taken a different approach with Claude, surfacing banners that direct users to Democracy Works’ TurboVote resources for current voting information rather than integrating that information directly into the chat experience.

• Self-harm. OpenAI now partners with the crisis-support organization ThroughLine to give ChatGPT users one-tap access to localized crisis lines built directly into model behavior, rather than leaving it entirely to the user to seek those services out separately. OpenAI has said this logic is designed to trigger when it sees mental or emotional distress signals in what someone writes or even what they are browsing with ChatGPT Atlas.

• Violence toward others. OpenAI has also said it routes conversations where a user appears to be planning to harm someone else to a specialized internal pipeline for human review, where trained staff can take action, including banning the account.

The infrastructure for detecting risk and routing it somewhere already exists. But in the context of violence prevention, LLMs appear to make these pathways stop at enforcement (a ban, a report) rather than opening a door (a crisis line, a disengagement resource), even though that very door — also referred to as an offramp — can deter a lot of violence. And don’t just take my word for it. Research from PERIL (the Polarization & Extremism Research & Innovation Lab at American University, which focuses on radicalization and extremist violence) backs this up: targeted violence is neither random nor inevitable. It follows identifiable pathways that often exist long before any law is broken — which means those pathways can be interrupted.

The ask for AI companies and regulators.

AI companies should build — and regulators should require — a version of the suicide-detection safeguard already being mandated in several states, extended explicitly and consistently to cover indicators of pathway-to-violence behavior, not just self-harm. Concretely, that means:

1. Detection informed by Behavioral Threat Assessment and Management (BTAM) best practices, not just keyword filters. Train models to recognize the same behavioral indicators NTAC and threat assessment professionals already use in schools and workplaces — leakage of intent, grievance fixation, escalating research into means, weapons access, and what NTAC’s own framework calls “desperation or despair” — not just explicit statements of self-harm.

2. A friction-and-referral response, not just a refusal or a ban. A flat refusal can just send someone to a less-safe platform, and an account ban (the current end state of OpenAI’s own harm-to-others pipeline) closes a door instead of opening one. A pause plus a real resource keeps the person reachable — and the underlying mechanism is already proven to work: when X tested pause-and-reconsider prompts on harmful replies, 34% of prompted users revised or didn’t send the replies, and people sent 11% fewer offensive replies afterward. That’s friction changing behavior on a mainstream platform at scale — the same logic just needs to be extended from harmful language to violence-pathway indicators.

3. One standard, applied consistently. The CCDH/CNN test proved this is a design choice, not a technical ceiling. Two platforms already refuse consistently where eight others don’t. There’s no excuse for that gap to persist industry-wide, and whether through regulation, industry standards, new laws, or some combination thereof, AI chatbot operators should have a robust standard for how to tackle these risks.

4. State and federal rules designed to prevent people from committing violence against others, not only themselves. Texas and Connecticut have already shown this language is workable. Other states should follow. The FTC’s open inquiry into seven major chatbot providers that build AI companions and their protections for children and teens is another way for the public — and for state legislators and regulators — to learn about the safeguards or lack thereof that companies have in place, and inform next steps.

5. Real accountability, not just disclosure. Washington’s and Oregon’s new chatbot laws already include a private right of action; Kentucky’s Attorney General has already sued a companion-AI company over child safety failures. Extending that same enforcement posture to violence-pathway detection would give this real teeth.

6. NTAC should weigh in directly. NTAC’s research provides a comprehensive blueprint for what “concerning indicators” should mean in an AI context, but it hasn’t yet issued public guidance connecting its threat-assessment framework to chatbots specifically. It should. The same way OpenAI already leans on outside clinical experts for its self-harm protocols, AI labs need a comparable authoritative partner for violence-pathway detection — even if that means simply clarifying that the same concerning indicators that apply in the real world, would also apply in the context of an AI chat.

7. A shared, cross-sector standard-setting effort. Civil society, companies, relevant experts, and government leaders should cooperate on something like what the Partnership on AI has convened for suicide prevention, but explicitly scoped to include targeted violence that individuals seek to inflict on others, not only self-harm. This should include AI labs, threat assessment professionals, mental health clinicians, civil rights lawyers, and NTAC, among others.

Much of this does not require new research. It requires applying evidence we already have — on threat pathways, friction design, off-ramps — to newer tech to understand how many existing tools can be translated to the AI context, and to identify where the gaps exist.

This is fixable and urgent.

Every week this problem goes unaddressed, more people are more likely to be having exactly the conversation with a chatbot that should trigger an intervention, with a system that isn’t built to recognize it. That’s not an inevitability. It’s a design and policy choice, and we can demand a different one.

If you’re at an AI lab, working on AI safeguards through legislation or otherwise, or funding related work, reach out at Hello@ThePublicServiceAlliance.com. We’d love to hear from you.

Take good care.

Isa Ulloa

PSA Founder & CEO

Subscribe now

Share

About PSA:

The Public Service Alliance (PSA) is the first nonpartisan platform built to empower America’s estimated 40 million current and former public servants — across all levels of government and political lines — to better and more affordably protect themselves and their families.

PSA operates three programs: (1) The PSA Marketplace, which provides streamlined access to free, discounted, and subsidized services that increase privacy, security, and professional and personal well-being; (2) The PSA Security Project, which produces original research and amplifies common-sense policy solutions to make public service safer; and (3) PSA Advisory Services, which pair former U.S. Secret Service, FBI, and other security experts with individuals and organizations to help them make better risk-mitigation decisions.

PSA’s work has been featured in the Associated Press, The Washington Post, Wired, and CBS Evening News, among other national media outlets. Learn more at PublicServiceAlliance.com.

One year ago this month, Minnesota state Representative Melissa Hortman and her husband Mark were murdered in their home by a politically motivated gunman. What investigators found in the suspect’s car should alarm everyone: a notebook with the handwritten names and home addresses of more than 45 state and federal officials, as well as a list of 11 data broker websites he used to research his targets for free or little cost. Court documents show that Hortman’s address was listed on her campaign website — a choice lawmakers sometimes make to signal that they’re accessible to their constituents. What this shows is that far too many public servants face risky exposure from multiple directions at once: data brokers, public records, and official websites. The system creates so many paths to their front door that closing one rarely closes the others. As a result, preventable tragic things happen.

Let’s start with data brokers because they often act with impunity and are a primary part of the backbone of infrastructure that’s fueling America’s public servant security crisis today. These brokers — which constitute a multi-billion-dollar industry — aggregate home addresses, phone numbers, family members’ names, and personal financial information, and they sell it very efficiently at scale, with virtually no accountability and negligible legal or regulatory restrictions. There is no real societal benefit to what they do. They just use people’s personal data to make money, regardless of the consequences.

Because of all of the anger, extremism, and dysfunction in our politics today, public servants are disproportionately exposed and at risk, especially women and people of color. Their names are often public by definition. Their roles make them identifiable. And the laws designed to protect consumers from data broker abuse largely weren’t written with public servants in mind. Congress did pass the Daniel Anderl Judicial Security and Privacy Act in 2022, which protects federal judges’ and certain other limited categories of public servants’ personal information from data broker resale, but similar protections for state lawmakers and other public servants have not followed in the ways they should. That’s a significant policy failure stemming from appalling legislative inaction on both sides.

The exposure gap is by design, it’s massive, and it doesn’t get enough attention. That’s why we analyzed the existing 19 state comprehensive consumer privacy laws and found them overwhelmingly weak in protecting public servants from data-related harms, as featured by Wired.

The gaps are not minor. They are structural. This is why data privacy tools that help people remove their personal information from a large volume of data broker sites on a recurring basis make up such a significant (and growing) portion of what PSA offers in our marketplace. It’s not a product category we stumbled into — it’s a deliberate response to what the evidence tells us about how public servants actually get targeted, reducing their risk of being doxxed, swatted, scammed, and more.

We want our users to increase friction between themselves and bad actors who might seek to target them and their families in the future. That’s why we scrutinize our marketplace partners to minimize the likelihood that our users are paying for services that don’t actually do what they say they do and we negotiate meaningful discounts so that these services are more affordable and accessible. We don’t take kickbacks or referral fees. We cut through the noise, saving our users time and money and empowering them to proactively elevate their privacy and security baselines.

The research backs up our approach. Rachel Kleinfeld’s landmark June 2026 Carnegie Endowment report on political violence in the United States, for example, specifically names doxxing protections as one of the most well-documented, evidence-based interventions for reducing threats against public servants. Not a nice-to-have. A documented reducer of violence.

That same report shows that nearly 90% of state legislators experienced threats or attacks between 2021 and 2024, and more than half of locally elected officials, including school board members, experienced threats or attacks from mid-2023 to 2024. And yet, most public servants today have no meaningful access to doxxing protections because the tools are fragmented, the laws are weak, and no one has built the infrastructure to close the gap at scale. That’s unacceptable. It puts the people who serve us at risk and makes the idea of serving in government unappealing to so many others who want to do good. That’s why PSA exists – to help close the gap.

To be clear: data privacy services — no matter how effective they are — are not a silver bullet. The underlying problem — a near-total absence of meaningful federal data privacy legislation and the serious structural gaps in state laws currently on the books — means that even the best data removal services are fighting a system that’s actively working against them, including because data brokers can re-aggregate and resell people’s information faster than it can be removed.

That’s why PSA is also raising national awareness of the nonpartisan policy reforms needed to fix this problem at the root. Our aforementioned report — “The Data-to-Violence Pipeline” — analyzes existing state-level comprehensive consumer privacy laws and recommends four key changes to state privacy regimes to make them stronger: (1) broadening coverage to all public servants; (2) requiring personal data takedowns within 10 days of a request to a data broker or government agency; (3) including private data sources and public records in the scope of protection; and (4) incorporating a broad private right of action to ensure accountability. This approach is essential to better protect the millions of Americans who have chosen to serve their communities and their country. And make no mistake: First Amendment principles and common-sense data privacy protections are not mutually exclusive.

At PSA, we don’t confuse awareness with action. Nor do we admire problems without building actual solutions. The data is out there. The solutions to reduce violence are too. That’s why we’re putting concrete tools in our users’ hands today, while elevating policy reforms to make public service in America safer at scale, including by working to make stronger data protections the new normal nationwide.

If you’re a current or former public servant -– at any level of government and across party lines -– PSA has your back. If you haven’t yet explored the data privacy and range of other tools in the PSA marketplace, we encourage you to do so. And if you’ve never served but know a current or former public servant — a colleague, neighbor, or family member -– share this free Substack post with them so they know they’re eligible for our support.

Check us out at PublicServiceAlliance.com and reach out with questions or feedback at Hello@ThePublicServiceAlliance.com. We’d love to hear from you.

Take good care,

Isa

Subscribe now

Share

About PSA:

The Public Service Alliance (PSA) is the first nonpartisan platform built to empower America’s estimated 40 million current and former public servants — across all levels of government and political lines — to better and more affordably protect themselves and their families.

PSA operates three programs: (1) The PSA Marketplace, which provides streamlined access to free, discounted, and subsidized services that increase privacy, security, and professional and personal well-being; (2) The PSA Security Project, which produces original research and amplifies common-sense policy solutions to make public service safer; and (3) PSA Advisory Services, which pair former U.S. Secret Service, FBI, and other security experts with individuals and organizations to help them make better risk-mitigation decisions.

PSA’s work has been featured in the Associated Press, The Washington Post, Wired, and CBS Evening News, among other national media outlets. Learn more at PublicServiceAlliance.com.

America is not just angrier than it was 10 years ago. We’re also way more stressed, isolated, and unhappy. The data is bleak. Over 60% of U.S. adults now report societal division as a significant source of stress, according to the APA’s 2025 Stress in America survey — and more than half report feeling isolated, left out, or lacking companionship often or some of the time. Things have gotten so bad that the former U.S. Surgeon General declared in 2023 that the epidemic of loneliness and isolation constitutes a public health crisis. And that’s not all. The United States ranked 23rd in the 2026 World Happiness Report — outside the top 20 for the second consecutive year. The bigger story though is young Americans: the United States ranks between 122nd and 133rd out of 136 countries in youth happiness, with life satisfaction among Americans under age 25 falling nearly a full point on a 0–10 scale over the past 20 years — one of the steepest declines anywhere in the world, at a time when young people in 85 of 136 countries are actually getting happier.

America is neurologically primed for escalation. Chronic exposure to threatening and emotionally charged rhetoric — in newsfeeds, social media comment sections, political ads, and more — activates the brain’s threat-detection systems in ways that are cumulative and compounding. Under sustained stress, the prefrontal cortex — responsible for judgment and impulse control — can get overridden. People perceive more stressful threats. They move faster. They post before they process. In other words, they react rather than respond. This is not an editorial. It’s neuroscience.1 And it’s happening at a massive scale, in no small part because online platforms are exploiting this vulnerability to drive highly profitable engagement cycles at the expense of public servants, millions of other Americans, and our democracy.

We Know How to Solve this Problem, We’re Choosing Not to Fund It, and the Opportunity Cost is Enormous.

PERIL — the Polarization & Extremism Research & Innovation Lab at American University — has spent years validating solutions for exactly this problem. Their work shows that targeted violence is neither random nor inevitable. It follows identifiable pathways that often exist long before any law is broken, which means they can be interrupted. Traditional law enforcement models typically wait until a line is crossed. A public health approach to this problem doesn’t. PERIL’s findings, alongside those of the broader political violence field, point to at least five interventions that work:

1. Behavioral Threat Assessment and Management (BTAM). The Secret Service’s National Threat Assessment Center (NTAC) has documented that targeted violence follows a pathway, meaning it’s rarely impulsive. Structured threat assessment programs can identify individuals on a pathway to violence before that violence occurs. Yet, these programs are dramatically underfunded at the state and local level, where most public servants actually work, as well as federally. Every jurisdiction should have a threat assessment team — and those that don’t should be required to establish one that adheres to best-in-class (perhaps even national) standards.

2. Bystander intervention. People in an at-risk individual’s immediate network are almost always aware that something is wrong before the individual attempts or commits an act of violence. Scalable bystander training — validated through multiple randomized controlled trials in campus and community settings — has shown meaningful results, and Princeton University’s Bridging Divides Initiative recently launched an updated community safety and de-escalation directory of programs doing exactly this work. However, many of these programs are also underfunded by both philanthropy and government — that should change.

3. Online friction and cognitive interrupt design. Small, well-placed friction — a prompt asking “are you sure?” after an algorithm detects potentially threatening language, for example — can meaningfully reduce follow-through. Not because it necessarily changes minds, but because it introduces a pause at exactly the moment judgment is most compromised. X tested this concept starting in 2020, and the results were striking: when prompted, 34% of people revised or chose not to send their reply at all, and after being prompted once, people composed an average of 11% fewer offensive replies going forward. The platform rolled it out broadly in 2021 — and then it quietly disappeared. What is abundantly obvious is that, today, most social media and other online platforms won’t independently choose to do the right thing if doing so might negatively impact their bottom lines. So absent responsible governance, two critical levers worth exploring are regulatory pressure and the risk of liability exposure. This, like everything else described in this piece, is therefore a policy choice — and failing to act is a choice, too.

4. Off-ramps matter. People on a pathway to violence are often reachable before they act — by family members, peers, trained social workers, or credible community messengers who’ve previously walked a similar path. Organizations like Life After Hate, for example, have demonstrated that off-ramping works when there is somewhere to go. Meanwhile, Moonshot’s data shows that people searching for extremist content online are significantly more likely to also click on mental health resources. That’s MAJOR. Let’s leverage it.

5. Lead by example. The rigorous research on this is clear: political leaders and others with high-profile influence play an outsized role in shaping whether violence is seen as acceptable. Silence can therefore be interpreted as permission. High-profile, bipartisan joint condemnations of political violence — like the moderated conversation between Sens. Curtis and Kelly, or the joint appearance by Govs. Cox and Shapiro — have measurable effects on reducing violent rhetoric. More leaders should be doing this loudly and consistently, regardless of who is the target of violence, using language that feels authentic and resonates outside of the Washington, DC bubble. For example, according to research conducted by PSA’s partner the Foundation for Public Service, using the phrase “public servants” rather than “government workers” resonates far more with the average American.

This Is Not an Accident. It’s Negligence. Let’s Do Better to Prevent Tragedy.

Let’s be real: the conditions that fuel political violence are well documented and exacerbated by platform design choices that prioritize engagement (and profit) over safety, laws that shield from meaningful accountability, and a decade of truly appalling legislative inaction across both parties.2

Responding to crisis after the fact is always more expensive than investing in preventing it in the first place. A peer-reviewed systematic review of 52 public health studies, for example, found a median return of 14 to 1 on prevention investments. Studies focused specifically on violence prevention — including by the National Academies of Sciences — point in the same direction. And yet we continue to fail to act upstream. That’s a choice. And it’s an unethical one that leaves taxpayers bearing the brunt of the cost.3

The point of this piece isn’t that we simply need to chill out, though honestly I think we would all agree that would be (really) nice.4 It’s instead that we have an evidence-based understanding of how escalation to violence works, knowledge of the validated set of interventions that can prevent it, and a persistent bipartisan failure to fund or mandate any of them at scale. This is the definition of a massive, unforced error.

Enough navel-gazing. The research-backed tools exist. The actionable playbooks are accessible. What’s missing now is the will to prioritize and invest in scalable prevention and crisis response where still needed. Funders and policymakers who are serious about this problem should be asking: Are we funding what works, investing in organizations with demonstrated track records of delivering actual (and measurable) results, and testing innovative solutions to close clear gaps? Are we willing to take measured risks, iterate rapidly (including by failing fast when it’s clear that’s the most responsible thing to do), and invest with time horizons long enough to reasonably assess impact?

At PSA, we don’t admire problems, so we’re doing two things simultaneously to meet this moment: (1) building practical, scalable tools — including the first-ever marketplace for America’s current and former public servants, providing them with faster and more affordable access to home security systems, data privacy services, experts like former Secret Service and FBI officials, and much more; and (2) producing original research to raise awareness of the common-sense, nonpartisan policy solutions that can make public service structurally safer over time. Both matter. Neither is sufficient alone.5

When fear of violence deters people from pursuing or remaining in public service, it makes America weaker. And the longer this continues, the harder it will be to fix. So, let’s not confuse process with progress and finally get to work.

If that energizes you, reach out at Hello@ThePublicServiceAlliance.com. We’d love to hear from you and explore what’s possible together.

We’ll be back in June with the next installment of The Brief. Until then, take good care.

Isa Ulloa
PSA Founder & CEO

Share

About PSA:

The Public Service Alliance (PSA) is the first nonpartisan platform built to empower America’s estimated 40 million current and former public servants — across all levels of government and political lines — to better and more affordably protect themselves and their families.

PSA operates three programs: (1) The PSA Marketplace, which provides streamlined access to free, discounted, and subsidized services that increase privacy, security, and professional and personal well-being; (2) The PSA Security Project, which produces original research and amplifies common-sense policy solutions to make public service safer; and (3) PSA Advisory Services, which pair former U.S. Secret Service, FBI, and other security experts with individuals and organizations to help them make better risk-mitigation decisions.

PSA’s work has been featured or cited in the Associated Press, The Washington Post, Wired, and CBS Evening News, among other national media outlets. Learn more atPublicServiceAlliance.com.

That’s the text I sent a friend and former coworker on Saturday night — after I received yet another news alert about political violence, this time about the shooting at the White House Correspondents’ Dinner, where I knew she likely was. That I had to send that text at all is crazy. That it’s far from the first time I and many others have sent a text like that should stop all of us cold.

Political violence of any kind is unacceptable, and it cannot be excused or justified by any grievance or ideology. Full stop. It doesn’t matter how much you like or dislike any particular political figure.

This incident made national headlines — as it should. But most threats against America’s public servants never do. And the response to the ones that do have become numbingly predictable: press releases often dressed up as outrage, calls for unity workshopped by out-of-touch consultants, and a news cycle that moves on within 48 hours. No sustained action. No structural change. Just the performance of concern — followed by silence.

That’s not enough. It’s never been enough. And the people (and their families) on the receiving end of these threats know it. They know it because the underlying conditions that make them the targets of violence — the toxic rhetoric, the readily available personal data, the online platforms that amplify threats without consequence, the laws and political incentives that enable all of it — haven’t meaningfully changed in a decade.

The harassment, intimidation, and violence targeting officials at every level of government — across party lines — have risen sharply since 2013. The school board member, county clerk, or local health official facing violent threats rarely makes the news cycle. They also rarely get access to the resources they need to protect themselves and their families. The data on this is alarming, and worth your time.

The money tells part of the story: federal campaigns and committees have spent over $100 million on security in the last decade alone, with security spending in the 2023-24 cycle at over five times what it was in the 2015-16 cycle. Candidates are now budgeting for bodyguards the way they once budgeted for yard signs. And by the way, the figure above is a conservative, minimum estimate and excludes ballooning spending from federal agencies such as the U.S. Secret Service — meaning the actual numbers at the federal level alone are much higher.

Let that sink in.

Protecting yourself while running for office — or simply doing your job in government — has become a line item. A cost of doing business. Something you plan for, budget for, and absorb. We have quietly accepted that the people who run our water systems, staff our emergency rooms, teach our kids, and enforce our laws should expect to be threatened as a condition of their service. That is a profound moral failure. Even more so because we have the tools to fix it — should we choose to do so.

But the threat to public service runs deeper than physical and digital safety. We’re failing public servants in myriad ways that compound over time — in pay, in mental health support, in basic job security — and the costs don’t disappear. They get deferred onto communities less equipped to absorb them, at higher total expense, with worse outcomes for everyone. The bottom line is that America is failing its public servants and it’s making us weaker.

Fear of violence is deterring people from running for office and otherwise engaging in public service. School board members are quitting in record numbers. World-class experts are leaving government rather than risking their families’ safety. This isn’t just a security crisis. It’s a democracy crisis. And in America, it should be unacceptable.

At PSA, we don’t admire problems — we work to solve them. That’s why we’re building the practical tools and raising awareness of the common-sense policy solutions to make public service meaningfully safer and more sustainable, so the 40 million Americans who serve — or have served — in government can more affordably access the resources they need and deserve, and so the structural conditions that have made public service in America more dangerous than ever can actually change.

Threats, harassment, and intimidation should never be the cost of doing business as a public servant in America. The people who show up every day to serve their communities and their country deserve to do that work without fear. That’s not a partisan position. It’s the basic standard every American should expect.

Enough thoughts and prayers. It’s time to act. If you want to be part of the solution, reach out at Hello@ThePublicServiceAlliance.com. We’d love to hear from you.

Take good care,
Isa Ulloa
PSA Founder & CEO

Subscribe now

Share

About PSA:

The Public Service Alliance (PSA) is the first nonpartisan platform built to empower America’s estimated 40 million current and former public servants — across all levels of government and political lines — to better and more affordably protect themselves and their families.

PSA operates three programs: (1) The PSA Marketplace, which provides streamlined access to free, discounted, and subsidized services that increase privacy, security, and professional and personal well-being; (2) The PSA Security Project, which produces original research and amplifies common-sense policy solutions to make public service safer; and (3) PSA Advisory Services, which pair former U.S. Secret Service, FBI, and other security experts with individuals and organizations to help them make better risk-mitigation decisions.

PSA’s work has been featured or cited in the Associated Press, The Washington Post, Wired, and CBS Evening News, among other national media outlets. Learn more at PublicServiceAlliance.com.

At the Public Service Alliance (PSA), we don’t believe in simply admiring problems. Last month, The Brief covered how we’re failing America’s public servants, why it’s making America weaker, and five actionable steps to make things better.

This month, we’re highlighting seven non-partisan nonprofits doing practical 501(c)(3)-eligible work to support America’s public servants and their families, helping hold together the connective tissue of our democracy in tangible ways.

Bookmark this list, share it widely, and consider donating via the links below.

Now, let’s jump in.

Share

1. Federal Employee Education & Assistance Fund (FEEA): FEEA.org

For nearly 40 years, FEEA has supported more than 60,000 families from almost every federal agency with no-interest emergency hardship loans, disaster relief grants, scholarships, tutoring, childcare subsidies, and more. Click here to donate to FEEA.

Think about what it means for every American when the people processing veterans’ benefits, running air traffic control, or inspecting our food supply are in financial crisis. FEEA keeps skilled, mission-driven people from having to choose between their work and their family’s financial survival – a choice no American should be forced to make, especially when that choice is forced by unnecessary political theatre.

FEEA provides confidential no-interest emergency loans, disaster relief grants, and merit-based college scholarships to federal employees and their families. They also offer free K-12 tutoring for lower-income federal families, and curate practical resources to help federal families navigate life in and out of work.

In times of government shutdowns or mass workforce disruptions, especially, FEEA steps up in real time. Click here to learn more and click here to donate to FEEA.

2. The STEPS Foundation: TheSteps-Foundation.org

STEPS supports Special Operations Forces spouses by providing them with the well-being resources they need and deserve, including to support the transition from military to civilian life. Click here to donate to The STEPS Foundation.

STEPS is helping close the profound gap in resources for the spouses who have also endured the significant sacrifices experienced by Special Operations Forces military servicemembers.

Each participant receives an individualized six-month support plan built around four cornerstones: mental health and wellness, physical health, community, and career and education. Programs for women include trauma-informed somatic coaching, meditation and mindfulness, nutrition and fitness, endocrine balancing, pelvic floor therapy, financial readiness counseling, ICF-certified performance coaching, and community connection calls. And 70% of STEPS service providers are spouses of current or former special operators themselves — experts with real proximity to the unique experiences of special operations families.

Public service is not a solo act, and neither is its cost. In 2023 alone, 98 military spouses died by suicide. Further, depression rates among military spouses are more than double those in the general population and a large share of active-duty spouses meet the criteria for Generalized Anxiety Disorder. These are not footnotes — they are urgent data points that reflect what years of deployments, relocations, isolation, and lack of support cost the people standing behind our service members.

Supporting the families behind public servants is not peripheral — it is an investment in the long-term stability of the institutions every American depends on. Click here to learn more and click here to donate to The STEPS Foundation.

3. Pirth: Pirth.org

Pirth provides preventative and rapid-response support for people facing any type of threat or harm online — while advancing systemic solutions that protect everyone. Click here to donate to Pirth.

Public servants are facing skyrocketing threats – online and in the real world. PSA partners with Pirth to support our users who are experiencing, or are at risk of, online threats and harms.

Through a secure, guided intake process, Pirth responds to reports of cyberbullying, doxing, extortion, hacking, harassment, image-based abuse, impersonation, and violent threats, providing curated resources and quick referrals, advocacy through their status as a Trusted Flagger with major online platforms, and ongoing case support. Every report submitted anonymously contributes to an aggregated dataset that drives evidence-based advocacy for better online safety policy — globally.

Through Pirth, you’re not just getting help, you’re contributing to structural solutions that make public service safer. Anyone facing online harm can use Pirth’s services, including current and former public servants.

In an era where violent rhetoric on social media platforms too often runs rampant, Pirth is a critical resource to protect against online threats, harassment, and abuse. Click here to learn more and click here to donate to Pirth.

4. Whistleblower Aid: WhistleblowerAid.org

Whistleblower Aid helps government employees and private-sector workers lawfully report allegations of misconduct and other wrongdoing on a nonpartisan basis. Click here to donate to Whistleblower Aid.

In today’s climate, people are risking everything to uphold the rule of law. That should not be the case in America.

Whistleblower Aid helps government employees – including active-duty servicemembers who may be facing illegal orders – and private-sector workers lawfully report misconduct on a pro bono basis. Their team provides pro bono legal representation, advocacy, communications strategy, and digital security for their clients. They also offer temporary rent or mortgage assistance and counseling for whistleblowers under extreme pressure.

Whistleblower Aid’s attorneys have represented some of the most consequential whistleblowers in modern American history, helping increase transparency and accountability in our democracy. Click here to reach their team confidentially and click here to donate to Whistleblower Aid.

5. FedsForward: FedsForward.org

FedsForward is unleashing America’s hidden talent by leveraging technology and community to support federal public servants navigating career moves across all sectors—including back into government. Click here to donate to FedsForward.

Federal employees with extraordinary expertise often struggle to communicate the portability of their skills across industries. Meanwhile, organizations across every sector would benefit from their unique expertise and capabilities — and government itself benefits when these experts return with enhanced cross-sector experience.

FedsForward blends AI-driven skills translation with peer networks, helping federal alumni connect, validate decisions, and maintain relationships that make transitions successful— including through their Resume Refresh tool, which transforms federal resumes into formats tailored for private industry.

Whether you’re an employer or a former fed, click here to learn more and click here to donate to FedsForward.

6. Work for America (WFA): WorkForAmerica.org

WFA empowers state and local governments to better recruit, hire, and inspire America’s public workforce. Through its Civic Match program, WFA connects mission-driven professionals, including former feds, with roles in state and local governments. Click here to learn more.

WFA’s Civic Match program connects former public servants and other mission-driven professionals with open roles across more than 250 state and local governments across the country. It’s free for both job seekers and hiring managers. And their platform includes a national resume bank, virtual and in-person hiring events, and dedicated talent leads who advocate on candidates’ behalf.

When federal positions are cut or left vacant, those skills don’t have to disappear from public service. They can move closer to home — into state agencies, city governments, and county offices that are often chronically understaffed and doing the most direct work in people’s daily lives. More than 12,000 public servants from all 50 states have joined WFA’s platform, and governors and mayors across the country are partnering with them to recruit top talent.

If you’re a public servant looking for your next chapter — at any level of government — explore Civic Match as a launchpad. Click here to learn more and get involved.

7. Foundation for Public Service: FoundationForPublicService.org

The Foundation for Public Service (FPS) develops and supports nonpartisan initiatives that strengthen public service across America. Click here to donate to FPS.

We cannot simply rely on the hope that talented people will choose unsustainable careers out of patriotism alone. Strong democratic institutions require hard-working, reliable, and honest people who feel safe and supported enough to serve, despite the inherent tradeoffs. That’s why FPS operates three programs:

First, it produces and amplifies free public education resources on issues facing current and former public servants, including in partnership with law firms providing pro bono support.

Second, it conducts rigorous research to identify how to communicate with the American public — across political lines and through trusted messengers — about why the safety of public servants should transcend politics, filling a critical gap in the political violence ecosystem.

And third, it convenes national bipartisan experts in closed-door sessions to identify systemic solutions that make public service safer, then amplifies and advances those findings through public education and key partnerships with nonpartisan organizations.

Click here to learn more and click here to donate to FPS to invest in a future for all Americans.

These nonprofits are helping talented, hard-working people choose public service and stay in it. They’re doing so in practical ways on a nonpartisan basis. That’s a cause every American has a stake in, regardless of party, regardless of which level of government you trust, and regardless of whether you’ve ever worked in government yourself.

Please consider donating via the links above and sharing this post widely.

You can also email us at Hello@ThePublicServiceAlliance.com to get involved with PSA and our partners.

We’ll be back with the next installment of The Brief next month. Until then, take good care.

Isa Ulloa

PSA Founder & CEO

Share

Learn more about how PSA is empowering America’s current and former public servants, across all levels of government and political lines, to better and more affordably protect themselves and their families at PublicServiceAlliance.com.

So, we’re working hard to fix it.

Public servants are facing skyrocketing threats. And for several reasons, including because they make public-sector salaries, the vast majority can’t afford to defend themselves. That’s why, last year, we launched the first marketplace for America’s nearly 40 million current and former public servants so they can save time and money on vetted services that help protect their privacy, security, and personal and professional well-being. We continue expanding the marketplace with new free, discounted, or subsidized services. You can check it out for free at PublicServiceAlliance.com.

We didn’t stop there. We also launched The PSA Security Project to raise national awareness of the threats facing public servants, across political lines, and help advance structural solutions to make public service safer.

As part of this work, we published today one of the most comprehensive reports to date on how existing privacy laws are failing America’s public servants. We encourage you to read our key findings on how we can advance common-sense data privacy laws without all-or-nothing approaches or violating First Amendment rights, as well as Wired’s coverage of our report.

And we also launched today, with this post, our free Substack – At Your Service – where we’ll share:

• A monthly update on the things that matter most to making public service in America safer via The Brief.

• Stories that we’re paying attention to and think you should be, too.

• Free privacy, security, and personal and professional well-being resources from our team.

• Highlights of what we’ve been working on across our Marketplace and The Security Project.

• And more… including coming-soon Community Notes to elevate the voices of America’s public servants.

All of this builds on the work we started last year, as outlined in our 2025 Impact Report. In less than one year from launch, PSA helped nearly 2,000 current and former public servants better protect themselves and their families, including by accessing at more affordable rates things like privacy and security services, lawyers, and well-being coaches. We also helped raise national awareness about not only the problem, but also the solution – including by partnering with The Impact Project to launch The Security Map – one of the most comprehensive datasets on threats to America’s public servants to date, as featured by CBS Evening News. Stay tuned for an expanded map soon.

We’ve got big plans for 2026 because a lot more work remains to be done to ensure that no American is forced to choose between serving their community and protecting their family.

We’re just getting started, we hope you’ll join us, and we’d love to hear from you. Please get in touch at Hello@ThePublicServiceAlliance.com.

Take good care,

Isabella Ulloa & The PSA Team