El 20 de noviembre de 2024, el Instituto de Seguridad de Inteligencia Artificial de los Estados Unidos del Instituto Nacional de Estándares y Tecnología (NIST) del Departamento de Comercio de los Estados Unidos anunció la formación del Grupo de Trabajo sobre los Riesgos de la IA para la Seguridad Nacional (TRAINS, por sus siglas en inglés). El grupo reúne a socios de todo el gobierno de los Estados Unidos para identificar, medir y gestionar las implicaciones emergentes de la tecnología de IA en rápida evolución para la seguridad nacional y la seguridad pública. Este anuncio se produce mientras Estados Unidos se prepara para albergar la primera reunión de la Red Internacional de Institutos de Seguridad de la IA en San Francisco.
El grupo de trabajo permitirá la investigación y prueba coordinadas de modelos avanzados de IA en dominios críticos de seguridad nacional y seguridad pública, como seguridad radiológica y nuclear, seguridad química y biológica, ciberseguridad, infraestructura crítica, capacidades militares convencionales y más.
Seguir leyendo contenido relacionado con esta política
Time Magazine
"Meanwhile,practical international collaboration on AI safety is advancing. Earlier in the week, the U.S. and U.K. AISIs shared preliminary findings from their pre-deployment evaluation of an advanced AI model—the upgraded version of Anthropic’s Claude 3.5 Sonnet. The evaluation focused on assessing the model’s biological and cyber capabilities, as well as its performance on software and development tasks, and the efficacy of the safeguards built into it to prevent the model from responding to harmful requests. Both the U.K. and U.S. AISIs found that these safeguards could be “routinely circumvented,” which they noted is “consistent with prior research on the vulnerability of other AI systems’ safeguards.”
Open Access Government
"Overall the task force’s plans are important in terms of global competition. As other nations, particularly China and Russia, invest heavily in AI, the U.S. is making sure that its AI innovations are safe, secure, and line up with national security goals.
Not only does the TRAINS task force hope to migrate risks, but it also looks to prevent adversaries from using AI to undermine U.S. security."
CSIS
"While the Seoul Statement is a good start for multilateralizing cooperation between AISIs, network members must now decide how to turn intent into action. At the November convening in San Francisco, they should strive to set the network’s goals, mechanisms, and international strategy in preparation for the AI Action Summit in February 2025. In doing so, they must ask tough questions, including about priorities, leadership, and membership."
Start writing here...