| Mobile Web

Search results for hallucination

AI & Enterprise

AI-driven security strategy sees generational shift as priorities change

Growing concern over cyberattacks that misuse advanced AI models is prompting calls for companies to reset security priorities. With AI able to find software vulnerabilities faster, the focus is shifting to minimising exposure time and adopting zero-trust approaches to contain spread even after a breach. Anthropic and Cloudflare warn that faster patching must be balanced with validation and testing. Governments and firms are also expanding AI-driven security programmes, partnerships and investments.

AI & Enterprise

Clion builds Seoul generative AI chatbot 2.0 using proprietary RAG

AI cloud services company Clion said on Tuesday it has completed Seoul Metropolitan Government\'s project to build a generative AI-based Chatbot 2.0. The project was pushed to carry out one of seven core tasks under Seoul\'s AI Vision, aimed at administrative innovation by converting public services to AI. Clion said it handled GPU infrastructure, data-based model tuning and RAG-based AI service development, applying its REX tool to improve chatbot response accuracy.

AI & Enterprise

DeepSeek-R1 hallucination rate quadruples versus V3 after stronger reasoning

China AI startup DeepSeek\'s latest reasoning model, DeepSeek-R1, recorded a much higher hallucination rate than its previous model, according to an evaluation by Vectara using the HHEM 2.1 benchmark. DeepSeek-R1 logged a 14.3 percent hallucination rate, about four times DeepSeek-V3\'s 3.9 percent. Vectara said cross-checks using Google\'s FACTS methodology showed a stronger tendency to add unsupported information, a concern as AI agent tokens automate crypto analysis and trading.