Alex Karp says AI could widen inequality—and rejects a wealth tax

Alex Karp says Palantir spent roughly three years building its early counterterrorism product without revenue. He sees that choice as a departure from the usual software playbook: the company pursued a difficult product intended both to identify threats and to protect civil liberties.

Walter Isaacson introduces Palantir as an attempt to apply techniques used to detect PayPal fraud to intelligence data after September 11. He asks Karp, the company’s co-founder and CEO, how powerful analytical systems can support security without sacrificing accountability.

Isaacson asks about Karp’s parents and childhood. Karp describes a complicated family and a messy divorce, and thinks growing up in that environment gave him confidence in his ability to survive while leaving him suspicious of established structures.

Asked about dyslexia, Karp says he kept it private for years. He describes reading as involving an extra translation step, but also thinks his experience encouraged him to approach problems without automatically adopting a conventional playbook.

Karp claims Palantir’s counterterrorism software helped prevent hundreds of attacks and that its products became important to Ukraine’s defense. He connects the decision to build it with a willingness to work for years without revenue; these are his characterizations of the company’s impact.

In Karp’s account, many software companies aim for a simple product that is difficult for customers to abandon and then rely heavily on sales. He contrasts that model with Palantir’s long investment in a product intended to meet two goals that different groups often treat as conflicting: security and civil liberties.

Isaacson suggests that being an outsider can be an advantage because it reduces pressure to fit in. Karp replies that people may stop trying to fit in because they expect the attempt to fail, and says Palantir’s culture and products reflect an outsider perspective.

Karp discusses his graduate work criticizing philosopher Martin Heidegger and his support for Hitler. He argues that apparently opposed political movements can share what he considers regressive assumptions, and says he eventually decided that he wanted to build a company rather than remain in academic philosophy.

He tells Isaacson that builders often combine an outsider’s perspective with aesthetic interests and business judgment. This is his explanation for successful technology companies, rather than a claim that technical ability is irrelevant.

Karp met Peter Thiel at Stanford Law School. According to Karp, Thiel later proposed bringing a method used against PayPal fraud into intelligence work; Karp found the idea compelling and joined the effort.

Karp describes his support for the American project while acknowledging slavery, failures of integration and insufficient attention to class. He argues that the country’s size, diversity and constitutional framework make it a distinctive political experiment.

He emphasizes freedom of expression, the right to bear arms and privacy. In his view, the first two are not sufficient without protections that allow people a private space to form unconventional or offensive ideas. Isaacson asks how that principle fits with Palantir’s data-gathering role.

Karp says European rules take privacy seriously in both commercial and government settings, while Americans may underestimate commercial overreach. He argues that people who are comfortable sharing data with companies may also underestimate how technology products shape their opinions.

Isaacson asks whether Palantir contributes to a surveillance culture and a loss of autonomy. Karp distinguishes monitoring conducted under a legal mandate from monitoring without one, and says Palantir’s products can enforce limits on retaining data when legal authorization expires.

Karp says the company’s tools are used against organized crime and terrorism under legal processes. He argues that rejecting all such monitoring can weaken public support for liberal institutions; Isaacson continues to ask about the broader consequences of data collection.

Asked whether AI and widespread data collection threaten autonomy, Karp says they can unless constrained. He regards AI as a raw material usable for good or ill and favors legislation preventing technology used on battlefields from being deployed domestically.

He argues that strong military capabilities are necessary for deterrence, but warns that a state pursuing absolute crime control can also suppress innocent people and political dissent. His argument is that security needs legal and civil-liberty boundaries.

Karp expects AI to worsen existing class inequalities even if it creates more or better jobs and some workers are retrained. He says technology owners could become many times wealthier while workers’ pay rises much more slowly.

Isaacson asks whether large technology-driven disparities justify putting more resources into shared public goods. Karp agrees that inequality needs a response and says he would accept higher taxation, but opposes a wealth tax, which he argues could damage technology companies and expand intrusive surveillance.

He sketches a trade-off involving estate taxation and identifying government waste alongside higher revenue. His proposal is that each unit of increased tax be paired with identifying half a unit of waste. He distinguishes reducing inequality from ensuring that additional resources actually improve poorer people’s lives.

Isaacson challenges the practical difficulty of valuing everyone’s assets. Karp replies that AI could make such assessments more feasible precisely by examining personal data, which is why he considers the surveillance implications central rather than incidental.

The discussion turns to data sovereignty and AI model providers. Karp claims some enterprise customers are angry about providers training models on valuable proprietary data and potentially competing with them. Isaacson asks how Palantir responds.

Karp says Palantir advises partners on contracts and interactions with closed and open models. He describes its ontology as a contextual framework that organizes raw information for use while serving as a protective layer, and argues that institutions and individuals should retain sovereignty over their data.

Isaacson asks whether more technology companies should support US military ambitions. Karp says Silicon Valley itself grew from cooperation among universities, government and capital, and criticizes a technology culture he sees as detached from national service.

Karp argues that a social bargain around wealth depends on people seeing value created for society. Isaacson asks whether technology created wealth through frivolous products. Karp says products may have value while an ordinary carpenter still perceives little material improvement compared with the gains of a small wealthy group.

He thinks AI could intensify that divide, particularly when technology leaders predict job losses while promising support funded by their own growing wealth. Isaacson connects this to a loss of national ambition; Karp says a shared ambition must offer people a believable improvement in their lives.

Asked about distrust of government, Karp points to powerful new technology, perceived mismanagement and a weak sense of national purpose. He says skepticism can be understandable, while warning against political extremes that exploit it.

Reflecting on conversations in Germany with former Nazis, Karp argues that charisma and the abuse of patriotism can attract people to extremist movements. He also says capitalism is better than available alternatives in his view, but does not solve every problem and can produce severe wealth and political-power imbalances.

Isaacson ends by asking what remains distinctive about people if machines surpass human cognitive ability. Karp appeals to a conception of human rights that values individuality and the human spirit beyond intelligence, and says those qualities should remain protected.