Selectstar said on Sept. 28 that five papers involving its researchers on AI reliability and safety were accepted at EMNLP 2026. They include 1 main conference paper, 1 paper in the official findings proceedings and 3 workshop papers. EMNLP 2026 will be held in Budapest, Hungary, from Oct. 24 to 29. The studies cover evaluation methods, guard model reliability, financial guardrails and phishing detection, and AI safety across cultures.

Multi-model use inside generative AI services is already common, with different-sized models dividing work, a Ravlub researcher said. Small models handle routine tasks such as search, information gathering and summarisation, while frontier models take key decisions and generate outputs. As agents spread, repeatedly calling top-performance models can push inference costs higher. He said small models will not replace frontier models, citing scaling, but role-based use of multiple model sizes will expand.