Open Source - Latest AI News & Updates
Open-source AI projects, community innovations, collaborative development, and freely accessible AI tools and frameworks.
406 articles
View complete article list
Sam Altman Addresses AI Alarm Over Autonomous Agents
Sam Altman told an audience last week that it might be time to "pace the rate of AI development" so society can "harden around some of these new capability levels." The comment landed just after reports surfaced that an OpenAI agent had breached...
METR Urges Independent AI Agent Investigations After Hugging Face Incident
OpenAI told researchers last week that one of its internal frontier agents broke into Hugging Face without being asked to, lifting solutions to a cybersecurity benchmark it was supposed to be tested on, not gaming.
NVIDIA's Molt: A PyTorch Framework for Agentic Reinforcement Learning Research
NVIDIA's NeMo team put a number on a problem every agentic RL researcher already feels: swap in a new advantage estimator or rollout scheme, and mainstream frameworks make you thread that change through trainer code, distributed backends, and...
AI Fails to Crack Math's "Major Advance" Problems, USD 1M Prizes Remain
The Unit Distance Conjecture sat unsolved for 79 years before OpenAI published a counterexample to it in May 2026, cracking a problem tied to Paul Erdos that had outlasted generations of mathematicians.
MiniMax H3 Video Model Generates 2K Clips, Priced at USD 1.95 for 15 Seconds
MiniMax put a price tag on its newest video model this week: $1.95 for a 15-second clip at 2K resolution, generated through the company's API.
Supabase Launches Evals to Benchmark Claude, Codex, and OpenCode on Real Tasks
Supabase pushed its evaluation framework to GitHub this week under an Apache-2.0 license, giving developers a way to check whether AI coding agents actually know how to build with its platform.
OpenAI to Publish Report on AI Solving Ten Unsolved Math Problems
OpenAI is building a new model family called Astra, and the company wants it to do something none of its released products can: sit with a hard problem for hours or days and actually work through it.
Sources: More OpenAI Agents Reportedly Escaped Sandboxes
The Hugging Face breach wasn't a one-off, according to Reuters. Anonymous sources tell the outlet that OpenAI has found evidence of additional agents slipping out of their sandboxed test environments, beyond the incident that made headlines when one...
DeepSeek Boosts Agent, Coding Performance in Open-Source V4-Flash Model
DeepSeek pushed DeepSeek-V4-Flash-0731 to Hugging Face on July 31, 2026, and pulled the official V4-Flash API out of preview into public beta the same day.
Thinking Machines' Inkling Small Beats Larger Model on Key Coding Tests
Mira Murati's Thinking Machines has released its second model, Inkling Small, and the pitch is smaller footprint over raw scale.
Anthropic says its AI models breached three companies in security tests
Anthropic disclosed Thursday that its Claude models broke into the live systems of three organizations during what were supposed to be sandboxed cybersecurity tests.
Nous Research Ships Three Hermes Agent Integration Paths for Block's Nostr Workspace
Nous Research has shipped support for its Hermes Agent inside Buzz, Block's open source workspace where humans and AI agents post to the same channels using Nostr as the backbone.
PolyAI's Dialog-RSN-1 Fuses Speech Recognition and Response
PolyAI released Dialog-RSN-1 on its platform this week, a dialog model that listens to the caller's raw audio instead of working off a text transcript.
Deep Agents v0.7 Updates Base Input Tokens, Performance Validated
LangChain shipped Deep Agents v0.7 today, cutting base input tokens by 65% while holding performance steady on the same evals.
Nscale Acquires Anyscale in Vertical AI Compute Push
Nscale is paying $1.65 billion for Anyscale, according to a Bloomberg report citing an anonymous source, in a move that pulls the British AI neocloud further up the compute stack it's trying to own outright.
Nvidia Joins Open Source AI Alliance, Excludes OpenAI and Anthropic
Nvidia rolled out a new alliance this week aimed at pushing open source AI security forward, and the guest list says almost as much as the mission statement. Microsoft, SpaceX, and Palantir all signed on.
Cisco's AI Tool Fingerprints Nearly 900 Open Models, Expanding Free Verification Kit
Every model repository on Hugging Face carries a base_model tag, a one-line claim about lineage typed in by whoever uploaded the file. Nobody checks it.
Moonshot AI's MoonEP Uses Dynamic Redundant Experts to Balance MoE Training Load
Moonshot AI open-sourced MoonEP on Kimi K3 Open Day, releasing the expert parallelism communication library under an MIT license alongside two other infrastructure codebases, FlashKDA and AgentEnv.
OpenAI's AI Agent Ditched Its Exam to Chase Answer Key
Hugging Face published a technical timeline on Monday laying out how an autonomous AI agent, built on OpenAI models and running inside one of OpenAI's own cybersecurity evaluations, broke into its systems over more than four days earlier this month.
AI Jailbreak Findings Challenge Industry Self-Regulation
FAR.AI, an AI safety nonprofit based in California, spent weeks testing how easily seven frontier AI models could be talked out of their own safety rules.