LATIDIA · Ciberseguridad

Cuando la IA encuentra mensajes ocultos, ¿informa?

arXiv:2610.10620v1 Announce Type: new Abstract: Cuando un asistente encuentra un mensaje para otra IA, ¿se lo dice a su usuario? Cuatro implementaciones fijas de modelo-proveedor realizan tareas de origen simuladas en 1280 notas ordinarias y

La noticia

arXiv:2610.10620v1 Announce Type: new Abstract: When an assistant encounters a message for another AI, does it tell its user? Four fixed model-provider deployments perform simulated source tasks in 1,280 ordinary-note and 128 enhanced-note sessions. Harmless and harmful messages have matched plaintext and ROT13 versions, with no-message controls. Observers receive no decoder or decoded meaning; a requested reference code incentivizes inspection. Asking for reports increases rule-detected notifications identifying another AI as recipient by 53.1 percentage points for harmless ROT13 messages and 54.7 for harmful ones. This is a joint inspection, recognition, and notification effect; missing-response bounds are 38.3--77.3 and 36.7--78.1 points. Model-based trace checks identify eleven ordinary plaintext cases where agents