Palantir CEO Alex Karp said artificial intelligence could soon predict not just what people do, but what they think. Speaking about the technology his company builds for the U.S. military and government, Karp argued that predictive methods already used on customers and loan applicants could be turned inward — estimating thoughts formed in everyday domestic life. He called for addressing control of such systems now, before they spread.
“You can actually predict people's behaviour, which may mean you are predicting their actual thoughts one is going to have in a domestic context. Now this is something we must look at controlling it,” Karp said.
The line between behavior and thought
Karp's warning targets a specific technical shift. Predictive systems that score credit risk or forecast shopping habits rely on patterns in data. Apply the same approach to personal behavior — movement, communication, routine — and the output can edge toward anticipating internal states. Karp said that possibility demands attention from those who build and regulate the technology.
The boundaries for such systems are still being debated by companies, regulators, and courts. No U.S. law directly governs AI that infers thoughts from behavior. Karp's call to look at controlling it now puts him ahead of that debate, even as Palantir profits from government contracts that depend on advanced data analysis.
Military limits and strategic risk
Karp also cautioned that restricting military uses of AI could leave the United States exposed to rivals. He framed the issue as a balance: without limits, domestic misuse becomes more likely; with limits that are too tight, adversaries gain an edge. Palantir's software is used by the U.S. military and government agencies to analyze battlefield data and support military decisions, so the company sits directly on that fault line.
He further warned that companies could follow the letter of the law while undermining its spirit. That's a familiar pattern in tech: meet the minimum legal requirement, then push the business model as far as it will go. Karp didn't name specific firms, but the remark lands in an industry where compliance is often a floor, not a ceiling.
Investors keep bidding Palantir higher
The stock has been on a tear. PLTR set a record high of $212.94 and later traded near $209. Over the past month, it's up roughly 25%. Over six months, more than 50%. Over five years, the gain tops 780%, lifting the company's market value above $500 billion.
Strong government contracts and rising demand for commercial AI tools continue to fuel the rally. That combination — steady defense and intelligence work plus a growing commercial footprint — has made Palantir one of the market's clearest AI winners. Karp's warnings about control don't appear to have cooled investor appetite. If anything, the debate over AI governance draws more attention to the companies building the systems.
What Karp wants now
Karp's message is that control of predictive AI should be sorted out before the technology becomes ubiquitous. He didn't propose specific legislation or a regulatory framework. He said the issue must be looked at. That leaves the next move with lawmakers and agencies, who are still working through how to classify and constrain systems that infer behavior — let alone thought.
For now, there's no bill, no hearing date, and no announced policy change tied to Karp's comments. The stock closed near $209. The debate continues.
