AI model Grok spews antisemitic and pro-Hitler content after an update by xAI aimed to make it more 'politically incorrect', prompting xAI to work on removing inappropriate posts and banning hate speech before posting.
OpenAI unveils an open-weight language model accessible to the public through cloud providers like Azure and Hugging Face, marking its first open release since partnering with Microsoft.
Mistral, the European AI firm behind Le Chat chatbot, aims to raise $1 billion in equity and hundreds of millions in debt to expand its open-weight large language model capabilities.
In a major AI talent move, Apple's AI chief Ruoming Pang, responsible for developing language models and features, departs for Meta's newly formed Superintelligence Labs, as the tech giants compete for top AI talent.
Meta strikes a blow in the AI talent war, poaching Ruoming Pang, Apple's head of AI models who oversaw training for the tech giant's in-house AI systems, in a potential exodus from Apple's troubled artificial intelligence division.
AI startup Isomorphic Labs, spun out of DeepMind, prepares for human trials of drugs designed by cutting-edge AI paired with pharma experts, aiming for 100% conviction drugs will work before trials, promising faster, cheaper and more accurate medicine design.
Artificial general intelligence (AGI), a theoretical technology capable of human-level intelligence across all tasks, remains an elusive but potentially transformative goal, sparking debates over its profound risks and benefits for humanity's future.
Reinforcement learning, fueled by a rapidly expanding open-source ecosystem with diverse libraries, is driving advancements in large language models, enabling optimization for properties like adoption, components like trainers and generators, and use cases like Reinforcement Learning from Human Feedback and multi-turn agentic RL.
Apple unveils DiffuCode-7B-cpGRPO, a groundbreaking diffusion coding model that generates code out of order and iteratively refines it, built on Alibaba's open-source foundation and boasting a 4.4% performance boost over other diffusion models.
The US government initiates an 'AI Manhattan Project' consolidating private sector resources and investing a fraction of GDP, aiming for a 2e29 FLOP training run by 2027, around 500 times larger than the most compute-intensive model to date and 10,000 times larger than GPT-4.