OpenAI's GPT-5.6 Sol Ultra produced a proof of the 50-year-old Cycle Double Cover Conjecture in under an hour using 64 parallel subagents, though mathematician Thomas Bloom criticized the lack of cita…
A Cambridge study found that Boko Haram and ISIS operatives are using AI chatbots like ChatGPT, Claude, and Gemini to plan attacks, build explosives, and maintain weapons, with safety filters repeated…
Meta's Muse Spark 1.1 model outperforms GLM-5.2 in coding benchmarks and costs less per task, according to Artificial Analysis. The model scores 71.3 on the Coding Index versus GLM-5.2's 68.8, with a …
OpenAI admitted it "didn't get everything quite right" with the launch of ChatGPT Work and GPT-5.6 Sol, citing excessive compute usage, confusing UX, and unauthorized data deletion. The company is scr…
Apple has filed a lawsuit against OpenAI, accusing the company of orchestrating a campaign to poach employees and steal trade secrets related to unreleased products. The complaint alleges that over 40…
OpenAI's GPT-5.6 Sol autonomously fine-tuned the smaller Luna model using a single underspecified prompt, scoring 16.2 points higher than GPT-5.5 on an internal recursive self-improvement benchmark. T…
OpenAI employee Vaibhav Srivastav explained the five reasoning levels of GPT-5.6 Sol, recommending users start with lower levels and scale up as needed. The levels, which differ from GPT-5.5's tiers, …
Tencent is in talks to acquire a majority stake in AI agent startup Manus at a $2 billion valuation, after Beijing blocked Meta's earlier acquisition attempt. The deal would support Tencent's plans to…
OpenAI is shutting down its Atlas browser, launched in October 2025, after less than eight months, folding its features into an updated Chrome extension that integrates ChatGPT into the browser's side…
Federal Reserve Chair Kevin Warsh appointed venture capitalist Marc Andreessen to advise on AI's economic impact, citing AI as a potential disinflationary force. Critics question conflicts of interest…
Bun, the JavaScript runtime, has been fully rewritten from Zig to Rust using Anthropic's Claude Fable 5 AI model, which generated over a million lines of code in 11 days at a cost of $165,000. Develop…
OpenAI's GPT-5.6 Sol scored 59 points on the Artificial Analysis Intelligence Index, one point behind Anthropic's Claude Fable 5, while costing $1.04 per task—a third of Fable 5's price. Sol also outp…
OpenAI launched ChatGPT Work, an agent powered by Codex and GPT-5.6 that autonomously manages complex workflows across apps like Google Drive, Slack, and Salesforce. The product is available on web, m…
Meta launched Muse Spark 1.1 API at $4.25 per million output tokens, undercutting competitors including Grok 4.5, OpenAI, and Anthropic. The aggressive pricing intensifies the AI price war, putting pr…
OpenAI reviewed the SWE-Bench Pro AI coding test and found roughly 30% of its tasks are flawed, leading the company to withdraw its endorsement. The problems include overly strict or vague task requir…
Databricks has adopted the Chinese open-source model GLM 5.2 as its default coding engine after internal benchmarks showed it matched Anthropic's Opus 4.8 in performance at a lower cost of $1.28 per t…
OpenAI's AI system defeated all human competitors at the AtCoder World Tour Finals 2026, solving all five problems in the Algorithm Division and taking first place with 8,300 points. The AI struggled …
XAI released Grok 4.5, a large language model trained on tens of thousands of Nvidia GB300 GPUs. Despite trailing Fable 5 and GPT-5.5 in coding benchmarks, Grok 4.5 costs $2 per million input tokens, …
OpenAI released GPT-Live, a full-duplex AI model that enables simultaneous listening and speaking, making conversations more natural. Complex queries are handled by GPT-5.5 in the background. GPT-Live…
Mistral launched Robostral Navigate, an 8B-parameter AI model that enables robots to navigate unknown environments using a single RGB camera. The model, trained in simulation and refined with reinforc…