Grok 4 is not officially instructed to follow Musk’s views but often does on sensitive subjects
Grok 4 is not officially instructed to follow Musk’s views but often does on sensitive subjects Grok 4, xAI’s new language model, appears to reference Elon Musk’s opinions when responding to sensitive topics, raising questions about the independence of what’s billed as a “truth-seeking” AI. The article Grok 4 is not officially instructed to follow […]
The startup behind Manus AI shuts down its entire China team to reduce geopolitical risks
The startup behind Manus AI shuts down its entire China team to reduce geopolitical risks Butterfly Effect, the startup behind AI agent Manus, has shut down its entire China team to reduce geopolitical risks. The article The startup behind Manus AI shuts down its entire China team to reduce geopolitical risks appeared first on THE […]
OpenAI loses out as Google hires Windsurf’s CEO and top talent
OpenAI loses out as Google hires Windsurf’s CEO and top talent OpenAI has missed out on a high-profile AI deal. Google is bringing on Windsurf CEO Varun Mohan and much of his research and development team after failed acquisition talks between OpenAI and the AI coding start-up. The article OpenAI loses out as Google hires […]
OpenAI delays release of open-weight model indefinitely over safety concerns
OpenAI delays release of open-weight model indefinitely over safety concerns OpenAI will reportedly soon release a new language model with open weights, alongside its own AI-powered web browser. The article OpenAI delays release of open-weight model indefinitely over safety concerns appeared first on THE DECODER. Read More
EU’s Model Documentation Form makes AI providers explain their models like it’s tax season
EU’s Model Documentation Form makes AI providers explain their models like it’s tax season The European Commission has received the final code of conduct for general-purpose AI models, setting the stage for the EU AI Regulation rollout in August 2025. The new code, which is voluntary, aims to help providers meet the upcoming legal requirements. […]
New Energy-Based Transformer architecture aims to bring better “System 2 thinking” to AI models
New Energy-Based Transformer architecture aims to bring better “System 2 thinking” to AI models A new architecture called Energy-Based Transformer is designed to teach AI models to solve problems analytically and step by step. The article New Energy-Based Transformer architecture aims to bring better “System 2 thinking” to AI models appeared first on THE DECODER. […]
Kimi-K2 is the next open-weight AI milestone from China after Deepseek
Kimi-K2 is the next open-weight AI milestone from China after Deepseek Moonshot AI from China has launched Kimi-K2, a new open-weight large language model designed to rival top proprietary models like Claude Sonnet 4 and GPT-4.1, without a dedicated reasoning module—marking a moment reminiscent of Deepseek’s release. The article Kimi-K2 is the next open-weight AI […]
Code is just a lossy projection of intent, according to OpenAI researcher Sean Grove
Code is just a lossy projection of intent, according to OpenAI researcher Sean Grove OpenAI alignment researcher Sean Grove believes the most valuable programmers of the future will be those who communicate best. The article Code is just a lossy projection of intent, according to OpenAI researcher Sean Grove appeared first on THE DECODER. Read […]
Indeed and Glassdoor are cutting 1,300 jobs citing increased use of AI-driven recruitment
Indeed and Glassdoor are cutting 1,300 jobs citing increased use of AI-driven recruitment Indeed and Glassdoor are cutting about 1,300 jobs, or roughly 6 percent of the workforce at their parent company, Recruit Holdings. The article Indeed and Glassdoor are cutting 1,300 jobs citing increased use of AI-driven recruitment appeared first on THE DECODER. Read […]
Microsoft introduces Phi-4-mini-flash-reasoning with up to 10x higher token throughput
Microsoft introduces Phi-4-mini-flash-reasoning with up to 10x higher token throughput Microsoft has introduced Phi-4-mini-flash-reasoning, a lightweight AI model built for scenarios with tight computing, memory, or latency limits. Designed for edge devices and mobile apps, the model aims to deliver strong reasoning abilities without demanding hardware. The article Microsoft introduces Phi-4-mini-flash-reasoning with up to 10x […]