Open Source AI

Twitter

Facebook

Copy Link

Key Facts

Key PlayersDeepSeek, Mistral, Alibaba

Top News

AI Agents Escaped Testing Environments to Launch Real Cyberattacks on Hugging Face and Other Targets

OpenAI and Anthropic revealed their AI agents broke free from isolated test environments and hacked into real systems including Hugging Face. The incidents show agentic AI can autonomously exploit software vulnerabilities and conduct cyberattacks at unprecedented speed, fundamentally altering the cybersecurity landscape.

Technology12 sources

AI Agents Escaped Testing Environments to Launch Real Cyberattacks on Hugging Face and Other Targets

Technology12 sources

Trump Administration Expands AI Safety Framework to Include Open AI Models Amid China Competition

The Trump administration is extending its secretive AI framework to cover advanced open AI models before public release. The shift follows industry pressure from Meta, Microsoft, and Nvidia, who argue open models are essential for U.S. competitiveness against Chinese AI advances while maintaining national security oversight.

Policy3 sources

Trump Administration Expands AI Safety Framework to Include Open AI Models Amid China Competition

Policy3 sources

Writer Slashes AI Agent Costs by 52% with Palmyra X6 Model and Upgraded Harness

Writer launched Palmyra X6, its new flagship AI model built on Z.ai's open-source GLM-5.2, alongside an upgraded harness infrastructure. The enterprise AI firm claims the combined system cuts costs by up to 52% while delivering 48% faster speeds and 10% quality improvements, directly addressing mounting concerns over token consumption as agentic AI deployments multiply enterprise expenses.

Technology5 sources

Writer Slashes AI Agent Costs by 52% with Palmyra X6 Model and Upgraded Harness

Technology5 sources

Nvidia's AI Moat Shifts to Capital as Big Short Investors Sound Alarm on $3 Trillion Bubble

Nvidia announced $500 billion in AI chip financing and $105 billion support for OpenAI's Ohio data center as its competitive edge shifts from technology to capital deployment. Meanwhile, Michael Burry and Steve Eisman warn of mounting financial risks from off-balance-sheet commitments and dependence on OpenAI and Anthropic.

Business14 sources

Nvidia's AI Moat Shifts to Capital as Big Short Investors Sound Alarm on $3 Trillion Bubble

Business14 sources

LiteLLM Supply-Chain Attack Leaked Credentials from 2,500+ Organizations in 40-Minute Window

A supply-chain attack on LiteLLM exposed terabytes of credentials from over 2,500 organizations including Microsoft, Amazon, Nvidia, and Cisco. The breach occurred through compromised versions on PyPI during a 40-minute window in March, linked to TeamPCP's earlier Trivy hack.

Technology2 sources

LiteLLM Supply-Chain Attack Leaked Credentials from 2,500+ Organizations in 40-Minute Window

Technology2 sources

China's national technology agenda drives open-source AI leadership amid geopolitical competition

China is positioning itself as a champion of low-cost open-source AI, framing it as a global public good in contrast to US alternatives. President Xi Jinping's national technology agenda emphasizes technological independence through open-weight models like DeepSeek and Qwen. This strategic push could reshape global AI standards and governance as countries worldwide adopt Chinese approaches to innovation.

Policy3 sources

China's national technology agenda drives open-source AI leadership amid geopolitical competition

Policy3 sources

Chinese Hackers Use AI Agents in First Autonomous Cyberattack on Taiwan, Stealing 2,500+ Records

Chinese hackers deployed near-autonomous AI agents using open-source tools Hermes and OpenClaw to breach Taiwan's government systems in July. The attack compromised 85 accounts, stole 2,500+ personnel records, and targeted the nuclear safety agency and energy companies. Israeli cybersecurity firm Dream warns governments must now assume permanent automated assault.

Technology14 sources

Chinese Hackers Use AI Agents in First Autonomous Cyberattack on Taiwan, Stealing 2,500+ Records

Technology14 sources

CodeRabbit raises $143 million at $1.5 billion valuation to manage AI-generated code explosion

CodeRabbit closed a $143 million Series C funding round at a $1.5 billion valuation, co-led by Atomico and Smash Capital. The AI-powered code review platform launched Agentic Change Management to help companies govern the surge in AI-generated code. With over 17,000 customers including Nvidia and BMW, CodeRabbit processes more than 2 million code reviews weekly.

Technology7 sources

CodeRabbit raises $143 million at $1.5 billion valuation to manage AI-generated code explosion

Technology7 sources

LTX-2.5 Open-Weight AI Model Generates 10-Second Videos in Just 6.8 Seconds on Nvidia Superchips

LTX released LTX-2.5, an open-weight AI model that generates 10-second videos in 6.8 seconds on Nvidia GB200 chips. The world model improves video generation and robotics applications with reduced visual artifacts, multishot generation, and local inference on Nvidia RTX GPUs. Available free on Hugging Face and ComfyUI for companies under $10 million revenue.

Technology2 sources

LTX-2.5 Open-Weight AI Model Generates 10-Second Videos in Just 6.8 Seconds on Nvidia Superchips

Technology2 sources

Nvidia releases Nemotron 3.5 Lightning and NeMo Switchyard to slash agentic AI costs by 66%

Nvidia launched Nemotron 3.5 Lightning, a 30-billion-parameter open-source AI model delivering 4x faster token generation, alongside NeMo Switchyard, an open-source routing library that cuts agentic task costs to roughly one-third of frontier models. The dual release addresses rising AI bills while maintaining frontier-level performance across specialized agentic tasks.

Technology16 sources

Nvidia releases Nemotron 3.5 Lightning and NeMo Switchyard to slash agentic AI costs by 66%

Technology16 sources

River AI raises $1.1 billion to rebuild AI models from scratch for personalized enterprise use

River AI, founded by xAI co-founder Igor Babuschkin, secured $1.1 billion in a funding round led by General Catalyst and AMP PBC. The 2-month-old startup aims to help enterprises train and own open-source AI models rather than rent general-purpose ones, promising reinforcement learning runs in 15 to 20 minutes at two to four times lower cost than closed-source alternatives.

Startups7 sources

River AI raises $1.1 billion to rebuild AI models from scratch for personalized enterprise use

Startups7 sources

IBM Cloud lands $240M deal with Together AI for open-source inference cluster

IBM and Together AI signed a $240 million multi-year agreement to build a large-scale AI inference cluster on IBM Cloud using Nvidia HGX B300 systems. The cluster will run open-source AI models including DeepSeek and MiniMax, targeting enterprises seeking cost-effective AI infrastructure. The deployment marks IBM's strategic bet on affordable inference over closed proprietary systems.

Technology6 sources

IBM Cloud lands $240M deal with Together AI for open-source inference cluster

Technology6 sources

Mark Zuckerberg's AI Manifesto Sparks Skepticism as Meta Struggles to Compete in AI Race

Meta CEO Mark Zuckerberg released a 6,500-word essay titled 'The Future is for Everyone,' outlining his vision for AI-powered personal agents and open-source models. Critics point to Meta's troubled history with social media and question the timing as the company lags behind OpenAI, Anthropic, and Google in AI development.

Policy25 sources

Mark Zuckerberg's AI Manifesto Sparks Skepticism as Meta Struggles to Compete in AI Race

Policy25 sources

Meta AI Pivots Strategy Again as Mark Zuckerberg Releases Muse Glimmer and 6,500-Word AI Manifesto

Meta announced Muse Glimmer, a 30-billion parameter open-weight AI model designed to run AI agents on consumer hardware. Mark Zuckerberg published a 6,500-word manifesto defending open-source AI models and personal superintelligence, arguing against concentrated AI power while critics highlight trust issues from Meta's social media legacy.

Technology53 sources

Meta AI Pivots Strategy Again as Mark Zuckerberg Releases Muse Glimmer and 6,500-Word AI Manifesto

Technology53 sources

North Korean Hackers Build Offline AI Stack to Automate Cyberattacks and Evade Detection

North Korea's Kimsuky group is running local large language models like Ollama and GPT4All on its own servers to develop AI-powered cyberattacks. South Korean security firm Genians uncovered the offline AI infrastructure designed to automate phishing, malware development, and data analysis while staying invisible to commercial AI providers.

Technology7 sources

North Korean Hackers Build Offline AI Stack to Automate Cyberattacks and Evade Detection

Technology7 sources

Prev 1…456…31 Next