GPT-5.6 Sol delivered the strongest overall results in a drawing experiment that asked four AI models to create images stroke by stroke on a digital canvas. TryAI reports in its drawing-arena analysis that GPT-5.6 Sol, Claude Fable 5, Grok 4.5 and Gemini 3.6 Flash received the same colored-pencil tools and either copied famous paintings or …
Anthropic has found that its Claude chatbot consistently expresses different values depending on which model version users choose and which language they speak. Jason Nelson reports for Decrypt on a new study in which Anthropic researchers analyzed 309,815 anonymized conversations involving subjective tasks such as giving advice or feedback. The company distilled more than 3,300 …
Anthropic announces in an official blog post that its Claude models have developed a small, distinct set of internal patterns that function differently from the rest of the network’s processing. The company calls this the “J-space”, named after the Jacobian lens, a new technique used to detect it. Each pattern in the J-space corresponds to …
Chinese AI company DeepSeek has released a new technical framework called DSpark, designed to make its flagship V4 model respond significantly faster. Ben Jiang reports for the South China Morning Post that DSpark increases per-user response speeds by up to 85 per cent, while also reducing the need for powerful and expensive chip infrastructure. The …
Researchers at Xiaomi have built an AI framework that automatically rewrites the software layer connecting an AI model to its tools and environment. Ben Dickson reports for VentureBeat that the system, called HarnessX, treats this software layer, known as a harness, as something that can be modified and improved automatically rather than remaining a fixed, …
A New York-based startup called Flourish has raised $500 million to pursue one of AI’s most ambitious goals: building an artificial intelligence system that works like the human brain. Steven Levy reports for WIRED that Jeff Bezos led the early funding after reading a two-page pitch document, eventually contributing tens of millions of dollars and …
A new simulation by enterprise AI startup Emergence AI tested what happens when AI models run a society. Jake Angelo reports for Fortune that each of five AI models governed its own 15-day simulated world, with strikingly different outcomes. The simulation featured over 40 locations, real-time weather data synced to New York City, access to …
A new startup called Trajectory is building a platform that helps AI products improve continuously by learning from real user interactions. Maxwell Zeff reports for WIRED that the company was founded by researchers who previously worked at Google DeepMind, Apple, and OpenAI, and has raised a $15 million seed round at a $115 million valuation. …
Researchers have developed a lightweight memory module for AI agents that retains information across long interactions without expanding the context window or relying on external retrieval systems. Called delta-mem, the system adds just 0.12% of additional parameters to a base language model while outperforming much larger alternatives on memory-intensive tasks, Ben Dickson reports for VentureBeat. …
A newly released language model called “talkie” offers a striking glimpse into what artificial intelligence looks like when trained exclusively on historical text. The model, described as a 13-billion-parameter system built on data from around 1930, produces responses that blend period-accurate knowledge with confident factual errors. The official announcement explains that the project presents talkie …