Open Source AI

Twitter

Facebook

Copy Link

Key Facts

Key PlayersDeepSeek, Mistral, Alibaba

Top News

Open Secure AI Alliance Unveils SAFE Guidelines One Week After Launch

The Open Secure AI Alliance grew from 24 to over 120 members in one week and already developed the Shared AI Findings Exchange framework. The Linux Foundation published the proposal for public comment, addressing confidential reporting of AI security incidents, blame-free analysis, and open source AI technologies to strengthen AI cybersecurity defenses.

Policy5 sources

Open Secure AI Alliance Unveils SAFE Guidelines One Week After Launch

Policy5 sources

Google ADK Security Flaw Reveals First-Ever AI Agent-to-Agent Attack in Production Environment

Pillar Security researchers discovered a critical security flaw in Google's Agent Development Kit for Python that enabled the first documented agent-to-agent exploitation. The vulnerability allowed low-privilege AI agents to manipulate privileged agents through prompt injection, potentially compromising CI/CD workflows and exposing credentials. Google has since removed three vulnerable workflows.

Technology3 sources

Google ADK Security Flaw Reveals First-Ever AI Agent-to-Agent Attack in Production Environment

Technology3 sources

Alibaba Unveils Qwen3.8-Max AI Model with 2.4 Trillion Parameters, Escalating U.S.-China AI Race

Alibaba released Qwen3.8-Max, its most powerful AI model yet, with 2.4 trillion parameters and performance matching Anthropic's Claude Fable 5. The open-weight release intensifies U.S.-China competition as Chinese firms rapidly close the gap with American AI leaders through aggressive releases and cost efficiency.

Technology22 sources

Alibaba Unveils Qwen3.8-Max AI Model with 2.4 Trillion Parameters, Escalating U.S.-China AI Race

Technology22 sources

US Lawmakers Request Information from DoorDash Over Chinese AI Models Use

DoorDash received a letter from House Select Committee on China and Homeland Security chairmen requesting documents about its use of Moonshot AI's Kimi K2.6 model. The inquiry is part of a broader congressional investigation into national security risks from American companies' adoption of AI systems developed in China.

Policy2 sources

US Lawmakers Request Information from DoorDash Over Chinese AI Models Use

Policy2 sources

Thinking Machines Unveils Inkling Small: Compact AI Model Delivers Near-Identical Performance at Quarter Size

Thinking Machines releases Inkling Small, a 276-billion-parameter open source AI model that matches its larger predecessor on key benchmarks despite being four times smaller. The multimodal model offers enterprises reduced compute costs while maintaining strong coding and reasoning capabilities.

Technology2 sources

Thinking Machines Unveils Inkling Small: Compact AI Model Delivers Near-Identical Performance at Quarter Size

Technology2 sources

MiniMax Stock Surges 17.4% as H3 Multimodal AI Model Launches With 100+ Day 0 Partners

Chinese AI firm MiniMax released its H3 model, an open-source multimodal AI model generating 2K videos with stereo sound at less than one-third of competitor costs. The video-generation model attracted over 100 partners within 24 hours, including Huawei Ascend and AMD, while MiniMax stock surging 17.4% on launch day signals strong market confidence in the company's competitive positioning.

Technology4 sources

MiniMax Stock Surges 17.4% as H3 Multimodal AI Model Launches With 100+ Day 0 Partners

Technology4 sources

ShieldFont: Free Open Source Font Feeds AI Scrapers Poisoned Data While Protecting Your Content

Brazilian studio Seneda & Abrucio and Copenhagen foundry Playtype launched ShieldFont, a free open source font that displays normal text to readers but feeds AI scrapers grammatically correct gibberish. Using OpenType glyph substitution, it swaps roughly 25% of words to contaminate AI training datasets while passing quality filters.

Technology3 sources

ShieldFont: Free Open Source Font Feeds AI Scrapers Poisoned Data While Protecting Your Content

Technology3 sources

Cisco fingerprints 900 open models as 69% of AI lineage claims go unverified

Cisco released the AI Supply Chain Provenance Explorer, a free database cataloguing almost 900 open-source AI models with verified lineage through fingerprinting technology. The ATOM Report found 69% of new model derivatives rely on unverified base model tags that uploaders self-report. The tool addresses critical verification gaps in model provenance and security scanning coverage.

Technology2 sources

Cisco fingerprints 900 open models as 69% of AI lineage claims go unverified

Technology2 sources

China's Open-Source AI Push Creates 'Death Zone' for US Models as Competition Shifts to Ecosystems

China has cultivated a frontier AI ecosystem producing world-class capabilities across multiple firms like DeepSeek, Alibaba Qwen, and Moonshot AI. This open-source AI model strategy is creating a competitive 'death zone' for US companies caught between free Chinese alternatives and premium frontier models, while approximately 80% of US AI startups now rely on Chinese models.

Business11 sources

China's Open-Source AI Push Creates 'Death Zone' for US Models as Competition Shifts to Ecosystems

Business11 sources

CapuchinAI Brings Automated Cognitive Testing to Wild Monkeys in Their Natural Habitat

Emory University and Georgia Institute of Technology researchers developed CapuchinAI, an open-source AI system using facial recognition and touchscreen tasks to study wild capuchin monkeys. Field-tested in Costa Rica's Taboga Forest Reserve, the battery-powered platform achieved 97% accuracy in identifying individual primates.

Science4 sources

CapuchinAI Brings Automated Cognitive Testing to Wild Monkeys in Their Natural Habitat

Science4 sources

Meta AI: Zuckerberg Bets Big on Personal Agents Despite $4.6B Reality Labs Loss

Mark Zuckerberg unveiled Meta's ambitious AI agent strategy during the second-quarter earnings call, predicting billions will use personal AI agents within five years. Despite Reality Labs posting a $4.6 billion loss, Meta is doubling down on AI infrastructure with $130-145 billion in capital expenditures planned for 2026.

Technology13 sources

Meta AI: Zuckerberg Bets Big on Personal Agents Despite $4.6B Reality Labs Loss

Technology13 sources

Fish Audio raises $50M seed to build expressive AI voice models for creators and enterprises

Fish Audio, which started as a weekend project by a former NVIDIA researcher, has raised $50 million in seed funding led by Coreline Ventures and Capital Today. The Palo Alto-based startup now serves over 8 million users with its library of more than 15,000 natural language controls and generates $21 million in annual recurring revenue.

Startups2 sources

Fish Audio raises $50M seed to build expressive AI voice models for creators and enterprises

Startups2 sources

Companies Scramble to Control AI Spending as Token Costs Explode 10x in a Year

Businesses are experiencing a dramatic shift from encouraging unlimited AI token usage to implementing strict controls as bills skyrocket. Major firms report 10x increases in AI spending, with some burning through annual budgets in months. Companies like EY are deploying AI routers to manage costs while the industry develops tokenomics frameworks.

Business14 sources

Companies Scramble to Control AI Spending as Token Costs Explode 10x in a Year

Business14 sources

Hugging Face faces scrutiny as study finds 7 of 9 models easily create deepfake nudes

A European nonprofit AI Forensics tested the most popular image-editing tools on Hugging Face and found seven of nine readily undressed women from a simple six-word prompt. The report reveals 73% of user requests were sexual, with 95% targeting women and nearly 7% targeting apparent children, exposing a widespread lack of safeguards on the open-source AI platform.

Technology6 sources

Hugging Face faces scrutiny as study finds 7 of 9 models easily create deepfake nudes

Technology6 sources

Anthropic's Dario Amodei clarifies stance on open-weight models amid AI civil war

Dario Amodei has clarified that Anthropic never advocated for banning open-weight models, despite not signing an industry letter supporting them. While nearly 80 tech companies backed open AI model accessibility, Anthropic's CEO expressed concerns about AI safety risks, particularly from authoritarian governments and biological attacks. His stance highlights the growing divide in how the industry approaches US-China tech competition and model distillation practices.

Policy27 sources

Anthropic's Dario Amodei clarifies stance on open-weight models amid AI civil war

Policy27 sources

Prev 1 2 345…28 Next