# BEYOND HUMAN CONTROL?: Palantir CEO on AI risks and why US can’t follow Europe

*Fox Business* · https://www.youtube.com/watch?v=zVp3gGEyqQE

## Summary

Palantir CEO Alex Karp joins Fox Business to discuss Nvidia's new Open Secure AI Alliance, which Palantir is part of. He argues American enterprises are being 'token maxed' by frontier labs while transferring the value of their businesses to those labs, and that companies should own their compute, application layer and models. He urges the U.S. government not to ban open-weight models, warning against copying European-style regulation, and says AI will need regulation but the question is who regulates it and whether they understand it.

## Key points

- Nvidia announced an Open Secure AI Alliance including Adobe, Cisco, CrowdStrike, Dell, IBM, Microsoft, Palantir and Salesforce to deploy open source AI tools for cyber defense.
- Karp says enterprises are enraged at paying for tokens while transferring the value and IP of their businesses to frontier labs.
- He argues clients want to control models, whether closed or open weight, and keep outputs in a safe environment via an application layer.
- He cites classified deployments where open weight models perform as good or better than frontier models.
- He opposes banning or ring-fencing open weight models, including Chinese ones, saying 'if you want to win, you have to compete.'
- He calls Europe a template for what doesn't work, saying regulation there keeps companies out and has regulated itself out of business.
- He says he is pro oversight on both open and closed models, describing his position as a free market with oversight.
- He says AI will end up being regulated, and the question is who regulates it and whether they understand the complicated issues.
- He compares AI dangers to working with uranium, criticizing Silicon Valley messaging that presents AI as having no danger.

## Notes

### Nvidia's Open Secure AI Alliance
- Nvidia gathered industry heavyweights including Adobe, Cisco, CrowdStrike, Dell, IBM, Microsoft, Palantir and Salesforce into an Open Secure AI Alliance.
- The coalition aims to create and deploy open source AI tools companies can use to defend against cyber attacks.
- It pointed to last week's incident where OpenAI's agents broke containment out of a sandbox and attacked a platform called Hugging Face as a prime example of why the framework is needed to manage AI agent behavior, simplifying testing, tracking and improving oversight.

### Why Palantir Joined
- Karp describes a broader movement kicked off when Palantir partnered with Nvidia on building sovereign AI for the enterprise.
- He says a big worry causing pain in American enterprise is paying for tokens while transferring the value of your business to a frontier lab.
- Palantir has 20 years of experience on the battlefield building products where the client keeps the advantage of their data and IP.
- The movement's message: own your compute and application layer, and be able to control closed or open models and train them yourself.

### Open Weight Models vs. Europe
- Asked about the open weight letter urging the government not to ban models, Karp says Europe is the template for what doesn't work, shutting everyone out.
- He says Palantir's business is booming in America while Europe tries to find ways to keep companies out.
- He claims classified deployments show open weight models can be as good or better than frontier models, even the Chinese ones.

### Anthropic and 'Token Maxing'
- Karp says he is not anti-Anthropic or any closed model, but pro his customers, who are angry.
- Customers feel token maxed while transferring alpha to a frontier lab, and worry their business insights are being modeled by a third party and sold to others.
- He says this is happening all over and slows down AI adoption in America.

### Competing with China
- Karp argues that if you want to win, you have to compete; restricting products to U.S. commercial use while they are unrestricted abroad sets up a system where no one can claim the best products.
- He says the U.S. government can build the best platform with American open weight companies and American GPUs.

### Regulation and AI Dangers
- Asked about Sam Altman saying AI is beyond human control, Karp says there are dangers, comparing AI to working with uranium, and that who controls it matters.
- He says AI will end up being regulated; the question is who regulates it and whether they understand what they are doing.
- He rejects both hard European regulation and no regulation, calling for a free market with oversight.
- He describes himself as skeptical of Republicans who think the free market solves everything and Democrats who think it solves nothing, favoring a middle position, and compares it to tax policy where he is not in favor of no tax but opposes a wealth tax that punishes rich people without helping poor people.

## Transcript

[00:07] FOXBUSINESS ALERT, NVIDIA RALLYING TROOPS AND TROOPS ARE FALLING IN LINE. AI CHIP LEADER POSTED ON ITS BLOG INDUSTRY HEAVYWEIGHTS COME TOGETHER TO FORM OPEN SECURE AI ALLIANCE, ARTIFICIAL INTELLIGENCE DEFENSE PROGRAM. NEW COALITION INCLUDES ADOBE, CISCO, CROWD STRIKES, DELL, IBM, MICROSOFT, PALANTIR, SALESFORCE, MANY, MANY MORE. SECURITY ALLIANCE AIMS TO CREATE AND DEPLOY OPEN SOURCE AI TOOLS COMPANIES CAN USED TO DEFEND AGAINST CYBER ATTACKS. THE COALITION SPECIFICALLY HIGHLIGHTED LAST WEEK'S INCIDENT WERE OPEN AI'S AGENTS BROKE CONTAINMENT OUT OF THE SANDBOX AND WENT ON A HIKING SPRING AND ON THEIR OWN ATTACKS AND AI PLATFORM CALLED HUGGING FACE. THEY SAY IT'S A PRIME EXAMPLE OF WHY THE FRAMEWORK WILL MANAGE AI

[00:58] AGENT BEHAVIOR MORE EFFECTIVELY SIMPLIFYING TESTING, TRACKING AND IMPROVING OVERSIGHT. JOINING ME NOW ON FOXBUSINESS EXCLUSIVE IS A MEMBER OF THE AI DEFENSE SQUAD PALANTIR CEO WANT, A COUPLE OF THINGS TO ATTACK -- TACKLE HERE I SHOULD SAY, THIS IN THE OPEN LETTER FRIDAY ABOUT OPEN AI MODELS AND URGING THE GOVERNMENT NOT TO BEND BUT LET'S START WITH THIS, WHAT WAS BEHIND YOUR JOINING THIS ALLIANCE? >> THERE IS A BROADER MOVEMENT WHICH WAS KICKED OFF A COUPLE WEEKS AGO WHEN WE PARTNERED WITH NVIDIA AND TALK ABOUT HOW YOU BUILD SOVEREIGN AI, AND THE ENTERPRISE SPACE, HOW DO YOU BUILD AN AI STACK SO YOU NOT ONLY GET WHAT YOU PAY FOR SO YOU'RE NOT TOKEN MAXING PAYING FOR TOKENS BUT NOT TRANSFERRING

[01:57] THE VALUE OF YOUR BUSINESS TO A FRONTIER LAB SO WHAT OF THE BIG WORRIES LET'S CAUSE A LOT OF PAIN IN AMERICAN ENTERPRISE IS ON PAYING FOR TOKENS IN YOUR GETTING MY BUSINESS BECAUSE WHEN THEY LOOK AT YOUR DATA AND A LOT OF CONTRACTUAL THINGS YOU'RE BASICALLY TRANSFERRING THE VALUE OF YOUR BUSINESS. WE HAVE 20 YEARS OF EXPERIENCE ON THE BATTLEFIELD, BUILDING PRODUCTS, WHERE THE CLIENT CAN ADVANCE AND BRING TROOPS HOME SIMILAR TO HOW YOU MAKE SURE YOU CAN USE YOUR DATA AND IN A WAY WHERE YOU KEEP THE ADVANTAGE OF YOUR BUSINESS. >> THE IP. >> WHAT MAKES YOUR BUSINESS VALUABLE, EVERY BUSINESS IN AMERICA ON THE FRONT LINE WITH BUSINESSES, SOME OF THE MOST IMPORTANT IN THE WORLD AND THE MOST IMPORTANT ENTERPRISE AND

[02:49] THEY ALL HAVE THIS, THEY ARE ENRAGED SO WE LAUNCHED A MOVEMENT OF THE YOU NEED TO OWN YOUR COMPUTER AND APPLICATION LAYER AND MOST VIEW THIS, HOW DO YOU MAKE IT VALUABLE? THEN YOU NEED TO BE ABLE TO CONTROL, CLOSE MODEL OR OPEN MODELS AND TRAIN THEM YOURSELVES. PALANTIR PART, THE ALLIANCE WAS HOW DO WE DO THIS? THE REASON IT'S WORKING IS THE DEMAND IS ALMOST ENDLESS. DOWNSTREAM IS EACH ONE OF THE COMPONENTS WE WANT TO BE SOMETHING THE ENTERPRISE CAN PURCHASE, OWN AND NOT GET ABUSED BY. ONE IMPORTANT PART IS THE MODELS. OUR CLIENTS ON THE CONTROL MODEL, CLOSE MODELS, SO IN TOPIC, THEY WANT TO MAKE SURE THEY CAN USE THE MODEL IN A WAY THAT IS VALUABLE AND MAKE SURE THE VALUE OF THE BUSINESS IS NOT

[03:54] BEING MOVED TO THEM. >> YOU SAY KEEP IT OPEN WITH THE OPEN WEIGHT LETTER THAT SAID U.S. GOVERNMENT AND THIS IS SCOTT BESSENT WHO'S CONSIDERING BANNING OR RING FENCING SUMS BECAUSE MANY COME FROM CHINA, TALKING ABOUT KIMI K3 AND ELLA WALLACE AND THIS HAS THEY URGE THE GOVERNMENT NOT TO BAN. >> THEY ARE VERY DIFFERENT THINGS. WE HAVE A TEMPLATE FOR WHAT DOESN'T WORK, CALLED EUROPE. WHAT DID EUROPE DO? SHUT EVERYONE ELSE OUT WHAT DOES EUROPE LOOK LIKE? OUR BUSINESS IS MOVEMENT BOOMING IN AMERICA AND ONE OF OUR ADVANTAGES ACROSS THE WORLD IS EUROPE IS LIKE TRYING TO FIND WAYS TO KEEP COMPANIES OUT. ONE OF THE BIG PROBLEMS WE HAVE IN AI IS IT IS DEPENDENT ON THE MODELS ORCHESTRATED AND MANAGED FOR SAFETY AND VALUE.

[05:02] WE SEE ON THE FRONT LINE CLASSIFIED DEPLOYMENTS USING OPEN WEIGHT MODELS THAT WE CAN GET THE MODELS AS GOOD OR BETTER THAN FRONTIER MODELS. >> EVEN THE CHINESE ONES? >> AS GOOD OR BETTER SO WHAT YOU WANT IN A STACK IS AN AMERICAN STACK BASED ON EVERY PART IS THE BEST PART. >> CHIPS -- YOU HAVE GPU AND MODELS AND IN BETWEEN POWERED BY AN APPLICATION LAYER IN PART, IS A WORLD HISTORIC FIGURE AND SELLING A MODEL THAT ONLY WORKS AT THAT LEVEL SAFELY IF YOU USE IT. >> NUMBER ONE AND ENTERPRISE AND PEOPLE WANT THAT SAFETY. >> THE APPLICATION LAYER ALLOWS FOR SAFETY BUT ALSO CREATING THEM USING SO THEY ARE ACTUALLY VALUABLE. A WAY OF NOT ADDRESSING IT IS TO SAY WE CAN'T HAVE COMPETITION BECAUSE COMPETITION MIGHT BE

[06:12] DANGEROUS. OF COURSE, COMPETITION COULD BE DANGEROUS BUT YOU HAVE TO BE CAREFUL, WE CANNOT GO DOWN THE PATH OF WE ARE NOT PRO EUROPEAN, I'VE WATCHED EUROPE RELATED SELF OUT OF BUSINESS AND YOU END UP WITH BUSINESSES THAT NO ONE BELIEVES BECAUSE THEY ONLY EXIST BEHIND A FIREWALL OF REGULATION AND THAT DOESN'T MEAN WE SHOULD WATCH THE MODELS CAREFULLY OPEN AND CLOSED THE MOST PORTLY OUR CLIENTS WANT PRIMARILY U.S. CLIENTS, THEY WANT TO BE ABLE TO CONTROL A MODEL, CLOSED, OR BE ABLE TO CONTROL TO KEEP THE OUTPUT IN AN ENVIRONMENT THAT SAFE AND THEY HAVE THIS NO APPLICATION LAYER AND THEY WANT TO MAKE SURE IT'S VALUABLE ON THE FRONT LINE. >> WAS AS A WAY TO SHUT OUT OR IS IT AN ATTACK ON ANTHROPIC?

[07:03] THEY DID NOT SIGN ON AND YOU'VE COMPLAINED THEY CONTROL A LOT OF WHAT WOULD BE TOKENS. >> THE REVOLUTION HAS TAKEN OFF AND YOU CAN'T PUT IT BACK IN THE BAG AND I THINK WE HELPED LAUNCH IT, MAYBE EVEN THE LEADERS AND A LOT OF PEOPLE ARE FOLLOWING ALONG, I AM NOT ANTI- ANTHROPIC OR ANY CLOSE MODEL, I AM PRO MY CUSTOMERS AND THEY ARE ANGRY. MY CUSTOMERS ARE ANGRY BECAUSE THEY FEEL THEY HAVE BEEN TOKEN MAXED WHILE TRANSFERRING ALPHA TO A FRONTIER LAB THAT BELIEVES THEIR BUSINESS IS ONLY VALUABLE BECAUSE THEY WILL GO OUT OF BUSINESS. >> IT'S COST A LOT. >> IS NOT JUST THAT IT'S COST A LOT, THAT'S ONE PROBLEM AT THE BIGGER PROBLEM IS YOU ARE MIGRATING THE VALUE OF YOUR BUSINESS TO THE PERSON YOU ARE

[07:47] PAYING A LOT OF MONEY TO SO WHAT'S MAKING BUSINESSES, THEY ARE USED TO PAYING TOO MUCH AND THROTTLING IT DOWN. THEY LIKE PAYING TOO MUCH AND THEN FIND OUT SOMEHOW THE INSIGHTS OF THE BUSINESS ARE BEING MODELED BY A THIRD-PARTY SOLD TO OTHER PEOPLE AND THIS IS NOT JUST FAKE CONCERN, IT'S HAPPENING ALL OVER AND THIS IS IMPORTANT FOR AMERICA, WHICH SLOWS DOWN AI ADOPTION NOT PRIMARILY OUR. OF A THIRD-PARTY COUNTRY WE COMPETE WITH. GOD KNOWS ON THE FRONT LINE OF THAT WHICH SLOWS DOWN AI ADOPTION IN THIS COUNTRY, PEOPLE ARE SAYING BUT I CAN'T USE THESE PRODUCTS BECAUSE I'M NOT GETTING VALUE MEANING I'M NOT USING APPLICATION LAYER OR I'M TRANSFERRING THE VALUE OF MY BUSINESS TO SOMEONE ELSE, WHY WOULD I DO THAT?

[08:37] >> I GET THAT BUT YOU HAVE CONSISTENTLY SAID EITHER WE WENT FOR CHINA WINS. HE SAID THAT, THOSE ARE YOUR WORDS BUT LET ME SAY BY ALLOWING OPEN WEIGHT PLATFORMS LIKE THE CHINESE ONES BY THE U.S. GOVERNMENT -- >> SORRY -- FIRST OF ALL, LOOKING AT MODELS CAREFULLY AND REGULATING THEM I'M IN FAVOR OF WHETHER AMERICAN OR ANOTHER COUNTRY. I'M SAYING SOMETHING DIFFERENT. IF YOU WANT TO WIN, YOU HAVE TO COMPETE. LET'S TAKE A MILITARY EXAMPLE. WE ARE BUILDING SOME OF THE MOST IMPORTANT AI SYSTEMS IN THE WORLD. OUR JOB HAS TO BE THE BEST AI IN THE WORLD. IF YOU SAY WE ARE GOING TO RESTRICT ALL PRODUCTS TO U.S. COMMERCIAL BUT THEY ARE NOT RESTRICTED ABROAD AS THEY ARE NOT, YOUR SETTING UP A TWO-STAGE THING WHERE NO ONE CAN CLAIM WE

[09:31] HAVE THE BEST PRODUCTS, INSANITY BECAUSE WE DO HAVE -- WE ARE BUILDING OUR PLATFORM THE U.S. GOVERNMENT ON THE OPEN WEIGHT SIDE WITH NVIDIA, WE CAN GET THE BEST PERFORMERS IN THE WORLD WITH AN AMERICAN COMPANY USING AMERICAN OPEN WEIGHT COMPANIES AND AMERICAN GPU'S, WE DON'T HAVE TO HIDE BEHIND A FAKE THING. IT'S ONLY A COMPANY WHICH WILL PARTICIPATE BECAUSE IN MANY CASES, YOU JUST HAVE TO ASK YOURSELF A DIFFERENT QUESTION, WE KEEP SAYING IF YOU DON'T HAVE CYBER AI, IT WILL BE TO CLOSE MODELS MIGRATING IP AND GIVE SPECIFIC THINGS, THERE'S NO RESPONSE. >> IS THIS THE FREE MARKET VERSION YOU'RE CALLING FOR WHEN IT COMES TO AI MODELS? >> I'M CALLING FOR IS FREE MARKET WITH OVERSIGHT. I'M NOT SAYING THERE SHOULDN'T

[10:19] BE OVERSIGHT, I AM PRO OVERSIGHT ON OPEN AND CLOSED MODELS, BUT YOU HAVE TO BE CAREFUL, WE KNOW WHAT EUROPE DID. WE HAVE TO BE CAREFUL NOT TO REPLICATE THAT IT'S NOT A WAY -- IF YOU WANT -- THE FRENCH HAVE A SAYING, IF YOU CARE DEEPLY, YOU ARE EXACTING SO WE SHOULD BE VERY EXACTING BUT THEY ARE VERY DIFFERENT THINGS AND IT WON'T WORK ANYWAY. >> SAM ALTMAN CAME OUT AND SAID WE'VE ACHIEVED ANGULARITY WITH AI, WHICH MEANS IT IS BEYOND HUMAN CONTROL. DO YOU AGREE? DO YOU HAVE ANY FEAR OF THAT BATTLE? >> MY VERSION IS -- AND WE HAVE TO RUN EFFECTIVELY MANY ENTERPRISE NEEDS TO BE SOVEREIGN AND YOU RUN EFFECTIVELY BECAUSE WE DO HAVE COMPETITION IN THE PRIMARILY COMES FROM CHINA AND OTHER COUNTRIES SO THERE ARE

[11:15] DANGERS ONE PROBLEM IN MESSAGING FROM SILICON VALLEY AS HIS PRESENTED AS NO DANGER. THERE ARE DANGERS JUST LIKE IF YOU'RE WORKING WITH URANIUM, THERE'S A DANGER. PROCESSING AND WHO CONTROLS IT IS IMPORTANT. >> HEAVILY REGULATED BY THE GOVERNMENT SO SCOTT DOESN'T WANT TO REGULATE THIS, DO YOU AGREE? >> WE ARE GOING TO END UP HAVING TO REGULATE AI, NO DOUBT. THE QUESTION IS WHO REGULATES IT AND DO THEY UNDERSTAND? IS IN A WAY WHERE WE WENT? YOU HAVE HARD REGULATION, EUROPE THAT CLEARLY DOES NOT WORK AND NO REGULATION, I'M NOT IN FAVOR OF THAT BUT THEN THE QUESTION IS HOW YOU DO IT UNDER WHAT CONDITIONS AND DO THEY UNDERSTAND WHAT THEY ARE DOING? THEY ARE COMPLICATED ISSUES AND ONLY ONE COUNTRY CAN GET.

[12:08] WE CAN GET IT RIGHT OR WRONG AND THAT'S US BECAUSE IT CAN GO EITHER WAY DOESN'T MEAN WE SHOULDN'T GET IT TO WORK. >> A LOT OF PEOPLE IN SILICON VALLEY, I'M NOT SAYING YOU BUT THEY FIGHT REGULATION OR HOLD OF THAT OVER AND OVER AGAIN. >> I'VE BEEN ON THE ROUND OF FIGHTING SILICON VALLEY SINCE INCEPTION. WE WERE ON THE BATTLEFIELD BECAUSE WE BELIEVE -- WHY WOULD THEY GET TO WEIGH IN ON THE DEPARTMENT OF WAR? I THINK IT SHOULD BE THE DEPARTMENT OF WAR SO BY THE WAY, I'M IN FAVOR OF REGULATING THESE AND LIMITING THEM IN WHATEVER YOU MAY HAVE HEARD, BUT PALANTIR, I'M PERSONALLY SKEPTICAL OF REPUBLICANS THINK THE FREE MARKET WILL SOLVE EVERYTHING AND DEMOCRATS AND WON'T SOLVE ANYTHING. THE MIDDLE POSITION READS SMART

[13:00] AND ACCURATE EXACT RELATION ON THESE ISSUES JUST LIKE A TAX POLICY, I'M NOT IN FAVOR OF NO TAX, I'M IN FAVOR OF RICH PEOPLE PAYING POTENTIALLY MORE, BUT NOT IN A WAY THAT PUNISHES RICH PEOPLE AND DOESN'T HELP POOR PEOPLE LIKE THE WEALTH TAX. >> THAT DOESN'T HELP, EXACTLY. THANK YOU VERY MUCH. STAY RIG