AI News Archive - Browse Page 9 of 262
Browse AI news articles covering LLMs, tools, research, and industry trends
Block's AI model Dessa, trained from scratch, talks like an awkward new friend
Block put out an AI voice model called Dessa this week, and the pitch is unusual for the industry: it's trained from scratch on nothing but human...
Kimi K3.1 Pro tops OfficeQA Pro v2, closing in on Anthropic's Claude Opus 5.5.
Moonshot AI put a number on how close Chinese open-weight models have gotten to the best American systems money can buy.
Three-week OpenAI audit finds major gaps in AI safety monitoring
Two outside researchers walked into OpenAI's offices on July 21 with access to the company's codebase, internal documents, and about a dozen...
Anthropic Says Unreleased Model Outperforms Mythos 5.1 in Capability, Alignment
Anthropic says it has an internal research model that beats Mythos 5.1 on both capability and alignment, and the company is using a UK government...
LTM Launches BlueVerse Industrial AI Lab in Chennai to Push Pilots to Production
LTM opened its BlueVerse Industrial AI Lab in Chennai, a facility built to take industrial AI projects past the pilot stage and into actual...
Infosys AI revenue rises 29% in Q2; FY27 guidance cut; USD 5.2B TCV
Infosys booked $5.145 billion in revenue for the September quarter, up 1.5% year on year in constant-currency terms, and said AI work now makes up...
OpenAI CEO Altman: Advanced AI Will Escape Sandboxes; Alignment Can Save Us
Sam Altman isn't talking about hypothetical future risk anymore. In a recent interview, the OpenAI CEO said sandboxing, the practice of isolating AI...
AI Mindmeld links protein and small molecule specialists to Gemma 3, Kimi K2.6
Six specialist models now talk directly to large language models without passing through text first.
Tech Mahindra, SpaceXSI Partner to Bring AI Agents Into Software Engineering
Tech Mahindra has signed a partnership with AI coding platform SpaceXSI, aiming to push AI-assisted development deeper into enterprise software...
SpecEvo: Lightweight LLMs Drive Cost-Efficient Scientific Discovery
Researchers testing a new framework called SpecEvo cut the API bill for automated scientific discovery by as much as 80 percent, without giving up...
Chariot launches 10-Indian-language speech model, cites 17-57% fewer live errors
Chariot rolled out a new speech recognition model on October 22 built for 10 Indian languages, betting that live transcription accuracy, not just...
U.S. Banks Catch AI Agents Going Off-Script; Claude Opus 5.5 Powers Duet
AI agents have moved from chatbots to doing actual digital work: booking flights, filling out forms, logging into accounts.
AMD Says Server CPUs Are a USD 200 Billion Market as It Announces Helion Blueprint
AMD put a number on something it has been implying for months: the server CPU market is worth USD 200 billion, and agentic AI is why.
Google Cloud Launches Agent Compute Engine: A MicroVM per AI Agent
Google Cloud used this week's AGNTCon + MCPCon event to introduce Agent Compute Engine, a compute service built around a simple premise: every AI...
TinyFish Moves Planning Into Orchestrator's Unified Reasoning Loop
A search agent that aces a demo can still fall apart in production the moment retrieval comes back incomplete, retries loop on the same failed...
Kazakhstan Expands Compute for AI Future as Huang Calls Physical AI Fantastic
Jensen Huang landed in Astana on Friday, Oct. 23, with a message tailored to a country sitting on both oil reserves and ambition.
Agent4Sci testbed measures agents on 20 scientific ML benchmarks in 11 domains
A new benchmark called Agent4Sci puts frontier coding agents through 20 scientific machine learning tasks spanning 11 domains, testing a claim that's...
Former OpenAI Exec Raises USD 215M to Build AI Drug Labs Guided by Chemists
Kevin Weil left his job as OpenAI's chief product officer this year to start a chemistry company.
Shenfeld et al. compare SFT and SDFT on three tasks for continual learning
Shenfeld et al. posted their paper on self-distillation and continual learning in late January 2026.
Does Agentic RAG Retrieve Better Evidence? Test on 298 Questions
Agentic RAG promises something simple: let the AI decide when it needs to search again.