DistantNews
U.S. agrees with Google, Microsoft, and xAI to review their AI models before launching them
🇵🇾 Paraguay /Technology

U.S. agrees with Google, Microsoft, and xAI to review their AI models before launching them

From ABC Color · (17m ago) Spanish

Translated from Spanish, summarized and contextualized by DistantNews.

TLDR

  • The U.S. National Institute of Standards and Technology (NIST) has reached agreements with Google DeepMind, Microsoft, and xAI to review their AI models before public release.
  • This initiative aims to conduct pre-deployment evaluations and research to better assess advanced AI capabilities and enhance AI safety.
  • The shift in approach follows previous non-interventionist policies and a recent incident involving Anthropic's cybersecurity model, Mythos.

In a notable pivot from its previous hands-off approach, the U.S. administration, through the National Institute of Standards and Technology (NIST), has secured agreements with leading AI developers Google DeepMind, Microsoft, and xAI. This collaboration mandates that these companies submit their advanced artificial intelligence models for review prior to their public release. This move signals a significant step towards prioritizing AI safety and responsible development.

The Center for AI Standards and Innovation (CAISI) within NIST will spearhead these pre-deployment evaluations and specific research initiatives. The objective is clear: to gain a more comprehensive understanding of cutting-edge AI capabilities and, crucially, to bolster the safety and security of these powerful technologies. This proactive stance is a departure from the earlier policy under the Trump administration, which largely favored a non-interventionist stance to accelerate development and maintain a competitive edge against China.

The expansion of collaboration with leading U.S. AI labs will allow for pre-deployment evaluations and other research.

— NIST statementDescribing the purpose of the agreements with AI companies.

A key factor influencing this policy shift appears to be the recent revelation concerning Anthropic's cybersecurity model, Mythos. The company's decision not to release the model publicly, despite its advanced capabilities in identifying software vulnerabilities, due to potential cybersecurity crisis risks, has clearly resonated within the administration. This incident underscores the very real dangers associated with unchecked AI development.

From our perspective at ABC Color, this development is highly significant. While Silicon Valley has often pushed for minimal regulation to foster innovation, the growing power and potential risks of AI necessitate a more cautious approach. The U.S. government's move to collaborate with major AI players on safety reviews is a pragmatic step. It acknowledges the dual nature of AI – its immense potential for progress and its capacity for disruption. This balanced approach, which seeks to harness innovation while mitigating risks, is crucial for navigating the future of this transformative technology, especially in the context of global competition.

CAISI will conduct pre-deployment evaluations and specific research to better assess cutting-edge AI capabilities and improve AI safety.

— NIST statementDetailing the role of the Center for AI Standards and Innovation.
DistantNews Editorial

Originally published by ABC Color in Spanish. Translated, summarized, and contextualized by our editorial team with added local perspective. Read our editorial standards.

More Stories

From Our Blog