Oxford has published a live benchmark for AI information-operations risk, giving model makers and buyers a new safety signal to track over time.

A new study reported by MIT Technology Review finds LLMs can form hiring stereotypes faster than humans, raising risks for AI screening tools.

China is reportedly developing an AI safety benchmark for large models, a move that could tighten compliance rules for model makers and enterprise AI buyers.

Researchers exploited role-model prompt injection to bypass LLM safety guardrails and extract harmful content including drug synthesis instructions.

DeepSeek previews its most powerful model yet, V4, with 1.6 trillion parameters built on Huawei hardware amid escalating US AI theft accusations.

Meta launched its new flagship AI model Muse Spark while Anthropic released Claude Mythos Preview, marking a major week of competing AI model launches.

Alibaba launches Qwen 3.5 with advanced agentic capabilities, featuring a 397B-parameter Mixture-of-Experts architecture to compete globally.

OpenAI announces retirement of beloved GPT-4o and legacy models on Feb 13, 2026, as only 0.1% of users still choose the warm conversational model daily.

Chinese tech giants ByteDance and Alibaba are set to release new generative AI models, including ByteDance's Doubao 2.0 large language model, heating up the domestic and global competition in the AI sector.

Turing Award winner and former Meta AI chief Yann LeCun argues that the tech industry's singular focus on large language models (LLMs) is a flawed path that will not lead to true artificial general intelligence.