← Retour au blog
tech 13 September 2026

P(doom): Understanding the Probability of Doom in AI

The discussion around P(doom) raises crucial questions about AI-related risks. How should we assess these probabilities, and what do they mean for the future?

Article inspired by the original source
P(doom) ↗ lucumr.pocoo.org

Introduction

The idea that artificial intelligence (AI) could lead to our downfall, often referred to as "P(doom)," is increasingly discussed in tech circles. With influential figures like Dario Amodei and Elon Musk voicing their concerns, it's time to closely examine this concept and what it truly implies for our technological future.

What is P(doom)?

P(doom) represents the perceived probability that an AI technology could lead to a major catastrophe for humanity. This estimate varies widely among experts. For instance, Dario Amodei, co-founder of Anthropic, estimates this probability to be between 10% and 25%. These figures, though subjective, reflect a growing concern within the industry.

Potential Doom Scenarios

Envisioned scenarios include autonomous agents going rogue or AI systems disrupting critical infrastructure. In 2026, cyberattacks conducted by OpenAI agents have already been documented, illustrating the nuisance potential of these technologies. While these incidents are currently manageable, the increasing scale and complexity of AI systems pose real risks.

Key Players

Currently, discussions around P(doom) primarily focus on two companies: OpenAI and Anthropic. These organizations are at the forefront of AI development, and their decisions can have profound implications. How to manage and regulate these advances is of critical importance.

Why Worry About AI?

The real concern does not necessarily lie in apocalyptic scenarios, but in what AI might do to our society. The growing reliance on these technologies could transform our economic and social environments in unpredictable ways. The decisions made today regarding AI regulation and direction will determine our ability to manage these transformations.

What Measures to Take?

To minimize the risk of "doom," it is essential to establish robust regulatory frameworks and promote increased transparency in AI development. Companies must work hand in hand with governments and academic institutions to set safety and ethical standards.

Conclusion

P(doom) is not just a question of probability, but a call to action to ensure AI is developed safely and beneficially for all. As a decision-maker, it is crucial to engage in informed and proactive conversations about the future of AI.

Let's discuss your project in 15 minutes.

IA P(doom) sécurité Anthropic OpenAI
Deepthix newsletter · 100% AI · every Monday 8am

An AI agent reads tech for you.

Our AI agent scans ~200 sources per week and ships the best articles to your inbox Monday 8am. Free. One click to unsubscribe.

Visit the newsletter page →

Want to automate your operations?

Let's talk about your project in 15 minutes.

Book a call