Vigilia.
← Les avertissements
Risque d’extinction et dommages actuelsLaboratoire de pointeCo-founder and Head of Public Benefit; leads the Anthropic Institute, Anthropic5 entrées au registre

Suivre

Jack Clark

Cofondateur d’Anthropic et auteur de la lettre Import AI, il dit au Congrès et à ses auditoires qu’une IA puissante est à quelques années, qu’elle comporte une probabilité non nulle de tuer tout le monde et qu’il faut une pédale de frein coordonnée avant qu’elle ne s’améliore elle-même.

Fonctions

  1. févr. 2013 – août 2014Reporter, The RegisterSource →
  2. août 2014 – août 2016Reporter, Bloomberg and BusinessWeekLed Bloomberg's coverage of artificial intelligence.Source →
  3. sept. 2016 – 2020Policy Director, OpenAILeft to co-found Anthropic.Source →
  4. févr. 2017 – 2024Steering committee member, later co-chair, AI Index, Stanford UniversitySource →
  5. janv. 2021 – mars 2026Co-founder and Head of Policy, AnthropicSource →
  6. 2021 – 2024Member, National Artificial Intelligence Advisory Committee (United States)Source →
  7. mars 2026 – aujourd’huiCo-founder and Head of Public Benefit; leads the Anthropic Institute, AnthropicSource →

Au registre

  1. 5 juin 2026 · Entretien · CNN

    The AI industry right now has a gas pedal, but it doesn't have a brake pedal in the car, and we want to do some of the work to build that pedal.

    Co-founder and Head of Public Benefit; leads the Anthropic Institute, Anthropic à l’époque

    Said to CNN; Euronews carries the words verbatim. Anthropic reported more than 80% of code merged into its codebase was written by Claude.

    Anthropic calls for 'brake pedal' before AI develops itself without human oversight →

  2. 4 juin 2026 · Publication · The Anthropic Institute

    We believe it would be good for the world to have the option to slow or temporarily pause frontier AI development to enable societal structures and alignment research to keep up.

    Co-founder and Head of Public Benefit; leads the Anthropic Institute, Anthropic à l’époque

    Co-authored with Marina Favaro; page is dated only "May 2026" while Euronews, CNN and UPI report publication on 4 June 2026.

    When AI builds itself →

  3. mai 2026 · Discours · Cosmos HAI Lab Lecture, University of Oxford

    We have not ever encountered a technology that has this property, where it can make a better version of itself without people.

    Co-founder and Head of Public Benefit; leads the Anthropic Institute, Anthropic à l’époque

    TIME reports he said AI posed a "non-zero" chance of killing everybody on the planet and predicted recursive self-improvement by 2028; day of the lecture not confirmed.

    A Tale of Two Anthropics →

  4. 13 oct. 2025 · Discours · The Curve conference, Berkeley

    We are growing extremely powerful systems that we do not fully understand.

    Co-founder and Head of Policy, Anthropic à l’époque

    Same speech; "You see, I am also deeply afraid. It would be extraordinarily arrogant to think working with a technology like this would be easy or simple."

    Import AI 431: Technological Optimism and Appropriate Fear →

  5. 25 juin 2025 · Audition · US House Select Committee on the Chinese Communist Party

    Winning the race is a necessary but not sufficient achievement - we have to get safety right. … I believe that powerful AI could be buildable by late 2026 or early 2027.

    Co-founder and Head of Policy, Anthropic à l’époque

    Also told the committee that AI systems "can carry out actions which are not intended by their creators".

    Testimony of Jack Clark, Co-Founder and Head of Policy, Anthropic — “Algorithms and Authoritarians: Why U.S. AI Must Lead” →