State Department Report Warns of AI Apocalypse, Suggests Limiting Compute Power Allowed for Training

Key Points:

  • Rapidly evolving AI poses a potential catastrophic risk to national security and humanity.
  • Experts warn of AI technology as an existential risk with potential extinction-level threats.
  • Recommendations include limiting computing power for AI training, government intervention, and criminalizing the sharing of AI model inner workings.


A recently commissioned report by the US State Department is warning of the “catastrophic” national security risks and potential threats to humanity posed by rapidly advancing AI technology. Titled “An Action Plan to Increase the Safety and Security of Advanced AI,” the report suggests urgent and decisive measures to mitigate these risks, including potentially limiting the computing power allocated to train AI systems. The report compares the potential destabilizing impact of advanced AI and artificial general intelligence (AGI) to the introduction of nuclear weapons, emphasizing the need for proactive government intervention.


Despite AI not yet reaching the level of AGI, where it could outperform humans intellectually, experts believe this milestone is imminent. The report, based on insights from over 200 experts in the field, emphasizes the gravity of the situation by echoing warnings from industry figures like Meta’s chief AI scientist Yann LeCun, Google’s head of AI Demis Hassabis, and former Google CEO Eric Schmidt. There is a growing concern among AI researchers that AI could lead to catastrophic outcomes, as indicated by recent surveys.


To prevent AI from posing an existential risk, the report proposes regulatory measures such as setting limits on computing power used to train AI models and requiring government approval for training AI beyond specified thresholds. Additionally, the report suggests making it a criminal offense to disclose the inner workings of powerful AI models, with the aim of preventing AI labs from losing control of their systems.


The report underscores the transformative potential of AI, from disease cures to scientific advancements, while cautioning against the attendant risks. Authors stress the need to enhance current safety and security measures, which they deem inadequate in relation to the imminent national security risks posed by AI.


Concerns regarding the regulation and governance of AI technologies are increasingly in the spotlight. While the European Union has taken steps to regulate AI, the US government’s response to the State Department report’s recommendations remains uncertain. Skeptics like Greg Allen, director of the Wadhwani Center for AI and Advanced Technologies, question the likelihood of US adoption of such stringent measures given the current regulatory landscape.


As the debate surrounding AI governance intensifies, the report serves as a stark reminder of the potential consequences of unchecked AI development. While it remains to be seen how governments will respond to these warnings, the discussion on balancing innovation with security in the realm of AI continues to evolve.



Prompt Engineering Guides



©2024 The Horizon