The conventional read on threats against AI executives is a story about dangerous fringe elements: a troubled man tailgating into Anthropic’s lobby, a Texas resident lobbing an incendiary device at Sam Altman’s home, dark-web actors amplifying violent rhetoric. The industry, in this telling, is an innocent bystander to irrational rage. I think that reading is almost exactly wrong — and the distinction matters enormously for what comes next.
The Thesis
I believe the surge in violent rhetoric and physical threats directed at AI companies is, at its core, a predictable consequence of a specific communications and deployment strategy — one in which executives spent years issuing stark warnings about AI’s power to eliminate jobs, then pivoted to triumphalism the moment venture capital dried up and the revenue pressure intensified, leaving ordinary workers with the anxiety but none of the promised upside. The anger is real, the grievances have a rational core, and the industry’s belated shift to armed guards and quieter branding is a symptom-management strategy, not a solution.
None of this excuses violence. Threatening to skin employees’ children or throw a Molotov cocktail at someone’s home is indefensible and criminal. But explaining is not excusing, and the technology industry has a long record of confusing the two when it suits them.
Supporting Argument: The Industry Built the Narrative That Is Now Consuming It
For several years, AI’s most prominent voices competed to issue the scariest possible predictions about the technology’s impact on work. The framing was almost always apocalyptic: entire professional categories wiped out, white-collar jobs automated faster than society could retrain, a looming displacement that would dwarf the Industrial Revolution. These warnings came from the same CEOs who now hire armed escorts and instruct employees to remove corporate logos before leaving unfamiliar neighborhoods.
The irony is not subtle. When Palantir’s Alex Karp told an AI and labor conference that fear of unemployment is feeding the backlash — that “when told your job is going to disappear, people go for the pitchfork” — he was diagnosing a dynamic his own industry engineered. Workers absorbed the job-killer rhetoric precisely because executives delivered it so forcefully and repeatedly. Now that same rhetoric has fermented into something uglier, and the response is not a reckoning with the messaging strategy, but a security budget line item.
The numbers tell the story clearly. In 2025, 38.1% of S&P 500 technology companies disclosed spending on executive protection, up from 26.8% in 2021. Palantir’s executive protection spending rose 150% in a single year to nearly $3 million. Oracle’s climbed 85.5% to $5.6 million, with disclosures noting “specific threats and safety concerns” around Larry Ellison’s residential security. Salesforce added roughly $1 million to its protection budget, reaching approximately $4 million. These are not rounding errors — they are institutional acknowledgments that something has structurally changed in the relationship between AI companies and the public.
Supporting Argument: The Grievances Are Economically Coherent, Even When the Expression Is Not
It is tempting to dismiss the threats as the work of the mentally unstable or the criminally inclined. Some certainly are. But the broader pattern of anger — passersby stopping outside an AI insurance startup’s café to shout about rising rents and stolen water, a laid-off Pinterest designer urging colleagues to “resist,” a Kansas City consultant reporting that clients feel they are personally training their AI replacement — reflects something coherent rather than chaotic.
A March 2025 Quinnipiac University survey of roughly 1,400 U.S. adults found that Americans concerned about AI outnumber those who are not by more than four to one. A majority — 55% — said they believed AI was doing more harm than good. These are not extremists. They are the median American, and their concerns about jobs, children’s well-being, and energy costs are grounded in real, observable phenomena. More than 200 economists and AI researchers have signed public warnings about job displacement, lending academic credibility to what workers are experiencing firsthand.
What makes this moment distinct from earlier technology panics — the dot-com bubble, early social media anxiety — is that the displacement is happening simultaneously with the triumphalism. Workers are being laid off in the same press cycles in which their former employers announce billion-dollar AI investment rounds. When Meta announced roughly 1,400 layoffs in Washington state during an AI pivot, Mark Zuckerberg’s yacht was photographed in Seattle waters. That juxtaposition is not lost on the people being displaced, and it closes the distance between abstract resentment and targeted anger faster than any prior technology transition managed.
Supporting Argument: The Security Response Is a Confession, Not a Solution
The industry’s practical response to the backlash — armed guards, logo-free clothing, quieter public messaging — is revealing precisely because of what it does not include. There is no announced policy commitment to retraining funds at scale. There is no industry-wide agreement on transition support tied to AI-driven layoffs. IBM’s public warnings about AI disrupting enterprise software and Satya Nadella’s cautions about enterprise AI trade-offs suggest that even within the industry, the honest accounting of AI’s costs is available — it is just not being translated into structural commitments to the workers on the receiving end.
Hiring armed guards is, in effect, a declaration that the industry expects the social friction to persist. JPT Security’s vice president of client relations, Dakota Dominguez, noted that tech CEOs “a few years ago, definitely did not have security,” and that the shift to armed protection is now routine budget planning. Anthropic has run round-the-clock security since 2024. One security employee noted his remit expanded from protecting CEO Dario Amodei alone to covering founders, C-suite executives, and their families globally. That is the security posture of an industry that has concluded this is the new normal, not a temporary aberration to be resolved through better public relations.
The digital threat data reinforces this. According to Liferaft, which monitors social media and the dark web for Fortune 100 clients, threats targeting AI executives and data centers rose sevenfold between late February and May 2025, before declining somewhat in June. “What has surprised me is how bad it’s gotten over such a short period of time,” said Jonathan Graff, Liferaft’s CEO. A seven-fold increase in fewer than three months is not a random fluctuation. It is a signal.
The Strongest Counterargument
The most serious objection to my thesis comes from civil libertarians and technology optimists, and it deserves a fair hearing. The argument runs as follows: AI companies did not cause deindustrialization, the housing crisis, wage stagnation, or the broader social frustration that predates the current AI wave. They are, in this view, convenient targets for diffuse anger that would find an outlet regardless. Blaming executives for warning honestly about disruption — and then blaming them for the reaction to those warnings — places an impossible obligation on corporate communicators to curate public emotion. Moreover, the specific individuals targeted, from Altman to Dario Amodei, had no personal role in any individual’s layoff. Holding them personally responsible for macroeconomic trends is a category error that the violent actors are making, not a legitimate political critique.
This argument has genuine force. It is true that AI is accelerating trends — automation, labor substitution, concentration of capital — that predate it by decades. The resentment has deep roots that no single industry created. AI spending is already warping macroeconomic signals in ways that make the distribution of gains and costs harder to see clearly, which compounds legitimate confusion about who bears responsibility.
But the counterargument fails on its own terms at the crucial margin. The specific acceleration of threats between February and May 2025 correlates with a specific period of high-profile AI-driven layoff announcements, concentrated in cities where AI companies are headquartered. The grievances are not purely diffuse. They are geographically and temporally concentrated around identifiable corporate actors. The industry’s own executives — Karp at the AI and labor conference — acknowledge that their messaging about job elimination is the direct proximate cause of the pitchfork dynamic. That is not a transfer of responsibility from bad actors to innocent companies. It is a recognition that words have consequences, and consequences have costs.
Why It Still Holds
The industry’s pivot — quieter messaging, armed protection, a new emphasis on AI’s benefits rather than its disruptive power — is not evidence that my argument is wrong. It is evidence that the industry has privately accepted it. You do not spend $3 million protecting a CEO from a threat you consider irrational. You do not instruct rank-and-file employees to hide their employer’s logo if you believe the public’s anger is misdirected and will naturally dissipate.
What the pivot does not do is address the underlying dynamic. AI-driven hiring tools are already screening white-collar candidates at scale, and AI background checks are scanning candidates’ full digital histories — further concentrating informational power with employers while workers have little recourse. The imbalance is structural, and a security budget cannot resolve a structural problem.
Bonnie Kate Wolf, the laid-off Pinterest designer, put it plainly: “You can’t go back to serfdom. It really feels like the people in power want to be kings. Historically, that doesn’t work out for kings.” That is not a threat. It is a warning, expressed with more historical literacy than most of the executives receiving it seem to have applied to their own situation.
Where This Ends Up
The most likely outcome is escalation managed through security spending until a serious incident forces a political response. Absent a material shift in how AI companies structure the economic relationship with displaced workers — transition support, profit-sharing mechanisms tied to productivity gains, serious retraining investment — the underlying pressure will not dissipate. Security budgets will grow. Executives will become more insulated. The companies that first move to announce credible, funded commitments to the workers their technology displaces will find the social temperature around them drops measurably — and gain a genuine competitive advantage in talent markets and regulatory goodwill.
The second-most-likely outcome is that a significant act of violence against an AI executive prompts federal intervention: hate-crime-style legislation targeting technology workers, expanded corporate espionage definitions, or FBI task forces focused on AI-adjacent threats. That would create a legal framework that protects executives while doing nothing for the workers whose anxiety is the fuel. If the industry wants to avoid that outcome — regulation that entrenches its power while ignoring the underlying grievance — it has a narrow window to act on the economics rather than the optics. The security guards are buying time. What happens with that time is a choice.
