- The CEO of Anthrope, Dario Amodei, estimates that a 25% probability AI leads to the catastrophe
- He still believes that it is worth investing the AI and that the benefits exceed the risks
- Their comments conform to the growing public and political conversations about the risks and the regulation of artificial intelligence.
One in four probabilities may seem quite good in some circumstances. They are much better probabilities than most casino games, for example. However, apparently it is unlikely enough for the anthropic CEO to seem carefree after setting the possibility that artificial intelligence leads to a disaster that ends society with 25%.
“I think there is a 25% chance that things are going very, very badly,” said Amodei happily at the Axios Ai + DC summit when asked about his (p) fatality, probability of fatality, belief around AI. But it is more concentrated in the “75% probability that things are going very well.”
By “really, very bad”, it does not mean that your phone is self -corrole “duck” to something worse. It refers to scenarios large enough to threaten social systems, existential risks, poorly used AI and fugitive results that could be catastrophic.
. @Jimvandehei asks the CEO of @anthropy @darioamodei what probability would give that AI end in disaster: “I think there is a 25% chance that things are going very, very badly.” #Axiososummit pic.twitter.com/9d7eqldyncSeptember 17, 2025
For an industry often soaked in utopian promise or reduced to the science fiction of Fearmongering, Amodei’s attitude about the probabilities of a apocalypse and why he is still moving forward with the technology that stood out.
Amodei is not just in expressing concern, but it is in a rare position. As CEO of the company behind Claude, it is not a passive observer. It is shaping the trajectory of this technology in real time. Your team is building the same systems whose potential and danger is weighing.
If someone told him that there was a probability of 1 in 4 that his car exploded every time he turned the key, he could start walking more. Amodei would apparently become a mechanic and check the car first before entering.
Ai doom
This is also not the only warning issued by Amodei about AI. It was warned before AI could eliminate half of all input level collar works and the alarm against US high -end chip exports to China sounded. That is what makes Amodei’s framing so useful. Recognize the risk, quantifies uncertainty, but leaves room for the agency.
On the other hand, the “75% probability of 75% amodei is going very well” is not optimism by itself. It implies the belief that AI could produce enormous benefits for all. It could lead to improved medicine, a more efficient manufacturing, and perhaps even strategies to address existential crises such as climate change (although a key element to solve this could be the energy required for AI models to work).
But 25% risk requires that these benefits be carefully built, with consideration for security measures and regulation. Because if the future is 75% bright and 25% broken, the question is: What are we going to do to keep the weight of the right side?
You may also like
