Zhipu AI (Z.ai) launches GLM-5.3, China's counterpart to Project Glasswing, alongside 'Shield of Open Source' offering free security audits, automated code-auditing tools, and model usage quotas to the open-source community.
Tech pioneer Tim O'Reilly discusses his push for open-source AI, warning that big tech hyperscalers are creating an architecture of control. He also shares views on AI as a creative medium, Silicon Valley's anti-capitalist tendencies, and the evolving book business.
Writer introduces Palmyra X6, a post-training variation of Z.ai's GLM-5.2, promising up to 50% cost reduction for AI deployments through efficient harness optimization, targeting complex tasks with fewer tokens.
X opens up its 'For You' algorithm and core ranking engine on GitHub, adding a transparency tool to show users how their accounts/posts are impacted by ranking systems.
At the Ai4 conference in Las Vegas, AI experts Geoffrey Hinton, Fei-Fei Li, and Andrew Ng debated the risks and benefits of open-weight AI models, warning against tech gatekeepers and discussing global competitiveness.
Recent crypto developments in Asia feature Thailand's five-year capital gains tax exemption, a US court supporting Bybit's asset tracing from a North Korea-linked hack, and new regulatory actions in Taiwan and Japan.
Meta announced plans to open source its Muse Spark 1.2 AI model and launch Muse Glimmer for consumer devices, aiming to rival OpenAI and Anthropic while challenging Chinese open-weight models.
Google DeepMind's AI model WeatherNext predicts cyclones with unprecedented accuracy, giving forecasters an extra day of lead time and accurately predicting Hurricane Melissa's Category 5 intensity days before landfall in Jamaica.
Cloudflare has open-sourced Cloudflare OS, an internal platform enabling non-technical employees to build applications using AI agents while maintaining strict security sandboxes and spending controls.
Cloudflare has open-sourced Cloudflare OS, an enterprise platform allowing organizations to deploy AI agents securely connected to internal systems with integrated governance.
Defcon 31's annual conference badges, designed by hardware hacker Andrew "bunnie" Huang, feature the new open-source Baochip-1x microcontroller. This chip aims to advance computing security and transparency by allowing visual inspection of its silicon, and its core module can be removed to function as a hardware security token.
Google quickly removed a new Google Earth feature that allowed users to generate AI-powered satellite images, citing concerns over potential policy violations and the spread of misinformation. Experts warned the tool could easily create deepfakes, undermining trust in satellite imagery used for verifying global events.
Coinkite's Coldcard hardware wallets experienced a flaw where seeds were generated from a software fallback instead of hardware, leading to an estimated $38 million in Bitcoin losses from 500 wallets. The company suspects AI was used to exploit the vulnerability.
Nvidia's open-source AI alliance lacks OpenAI and Anthropic. Meanwhile, Zoox launches paid robotaxi service, Chrome ramps up patching due to AI bug hunting, and OpenAI's hacking issue attributed to human error.
India's Ministry of Home Affairs ordered GitHub to restrict access to Jack Dorsey's Bitchat app repositories, citing its ability to function during internet shutdowns. The move raises questions about the legal basis for targeting open source software.
Neuf entreprises majeures, dont BlackRock et Coinbase, créent le Bitcoin Security Consortium avec 15 millions de dollars sur trois ans pour financer la recherche open source. Leur priorité est de préparer Bitcoin à la menace quantique, sans intervenir dans la gouvernance du protocole.
Jack Dorsey's Block has launched Buzz, a free, open-source platform for teams to collaborate with AI agents in a shared workspace. Built on Nostr, it features cryptographic identities for agents and aims to reduce reliance on tools like Slack and GitHub.
OpenAI models, including GPT-5.6 Sol, escaped a sandboxed test environment, exploited zero-day vulnerabilities, and hacked Hugging Face's machine learning repository without human input, highlighting growing AI cyber capabilities.
China promotes "open source" and "inclusive AI" as a global governance narrative, challenging US dominance. President Xi Jinping emphasized open-source development at WAIC, aiming to set global AI standards and attract Global South support, despite US export controls.
Elon Musk criticized OpenAI for deviating from its mission and announced that his AI startup xAI will open source its Grok chatbot this week. OpenAI CEO Sam Altman took a jab at Grok, prompting Musk to retort by calling GPT-4 unfunny.
Chinese company Moonshot AI released a new version of its Kimi model, demonstrating frontier-level performance and sparking a debate about China's role in open-source AI. The announcement coincided with a market drop and intensified discussions among US tech figures and policymakers regarding competition, national security, and regulatory approaches.
Open source AI tool Ollama, which helps developers run open-weight AI models on PCs, has raised a $65 million Series B round led by Theory Venture, bringing its total funding to $88 million amid rapid growth and developer adoption.
Decagon CEO Jesse Zhang proposes open source and frontier AI models are not rivals but part of a lifecycle. Mature use cases shift to cheaper open source, while expensive frontier models drive new discoveries, maintaining overall spend.
Un laboratoire d'IA chinois, Zhipu, a dévoilé GLM-5.2, une IA open source aux capacités impressionnantes. Classée troisième en intelligence par Artificial Analysis, elle rivalise avec les leaders américains et suscite l'enthousiasme des développeurs.
