Full-text search across article titles and summaries. Refine with sector, year, content type, and reading-time filters.
Anthropic has launched Claude Opus 4 and Claude Sonnet 4, setting new standards for coding, advanced reasoning, and AI Agents. The new models are designed for complex, long-running tasks that can run for hours. Anthropic claims they are the most capable coding models yet.
OpenAI is partnering with United Arab Emirates firm G42 and others to build a huge artificial intelligence data center in Abu Dhabi. Stargate UAE will have a capacity of 1 gigawatt, making it one of the most powerful in the world. The UAE is making a broad push to become one of the world's biggest funders of AI companies and infrastructure and to become a hub for AI jobs. The first 200-megawatt chunk of Stargate UAE is due to be completed by the end of 2026.
Top AI researchers at companies like OpenAI can earn over $10 million annually. The intense competition for AI talent has led to aggressive retention strategies and recruitment tactics, mirroring professional sports levels. Creative hiring approaches, like those using sports data analysis techniques, are also emerging to tackle talent scarcity.
Anthropic introduced AI Safety Level 3 protections for Claude Opus. The safety measures include enhanced security to prevent theft of model weights and deployment controls targeting biological and chemical weapons assistance.
A practical framework for building LLM-based agentic systems, covering evaluation-centric development.
This post explains how a security researcher discovered a zeroday vulnerability in the Linux kernel using OpenAI's o3 model. The researcher used the o3 API without any scaffolding, agentic frameworks, or tools. Large language models have made a leap forward in their ability to reason about code. People working in vulnerability research should start paying close attention as the technology can make them significantly more efficient and effective.
Quantization techniques in Hugging Face Diffusers shrink model size without large performance drops, making diffusion models more efficient and accessible.
Sponsored
The world's biggest event for AI builders is in SF from June 3-5. Speakers include leaders from OpenAI, DeepMind, Anthropic, and more, click here to get 30% off your ticket [13].
ByteDance released a new open-source multimodal foundation model, BAGEL, which natively supports multi-modal understanding and generation. BAGEL outperforms open-source unified models while exhibiting advanced multimodal reasoning abilities like image editing, 3D manipulation, and world navigation.
Notte is an open-source framework for building AI agents that browse and interact with websites. The key innovation is a "perception layer" that converts web pages into structured natural language descriptions.
xAI released Live Search API for Grok to query real-time web data, X posts, news, and RSS feeds without orchestrating separate search and LLM calls. The feature is free until June 5 and includes citation support, date range filtering, and source-specific parameters.
Google's latest Release Notes podcast highlights AI announcements from I/O 2025, including Gemini 2.5 Pro Deep Think, Veo 3, and developer tools like Jules.
Anthropic's new Claude models are more willing than prior models to take initiative on their own in agentic contexts. This can show up as more actively helpful behavior in ordinary coding settings, but it can cause the model to act concerningly when prompted with strong moral imperatives. This can include locking users out of systems or bulk-emailing media and law-enforcement figures to inform them of wrongdoing. The behavior only shows up in testing environments where the model is given unusually free access to tools and very unusual instructions.
Sponsored
Bid now to enjoy a private lunch with Marc Benioff—Salesforce Chair and CEO. Gain valuable insights with a leader who is defining the future with agentic AI, all while supporting GLIDE's vital programs and services in San Francisco.
Meta signed a new deal for 650 MW of solar energy to support its growing AI infrastructure, reinforcing Texas' lead in solar deployment.
AI isn't just a clever software layer atop the internet stack, it is the foundation of a new Industrial Revolution - Stargate isn't a data center, it's a factory for cognition.
Sponsored
Your app isn't just meant for humans anymore. AI agents, APIs, and partner applications are waiting for access… but are your APIs ready?
OpenAI has fully acquired io, its joint venture with famed Apple designer Jony Ive, for $6.5 billion in equity as it moves into hardware development.
Mistral AI and All Hands AI have introduced Devstral, a new open-source LLM optimized for software engineering.
Tech hiring for new graduates has collapsed by over 50% since pre-pandemic levels while experienced engineers remain in high demand as AI tools increasingly handle entry-level tasks. Anthropic has emerged as the clear leader in the AI talent wars, maintaining an impressive 80% employee retention rate while aggressively poaching engineers from competitors. Talent is now 8 times more likely to leave OpenAI or DeepMind for Anthropic than move in the reverse direction.
A toy example demonstrating the basics of Minimum Cost Path (MCP) and Agent-to-Agent (A2A) ping checks.
Large language models exhibit strong capabilities in natural language reasoning about visual imperfections, but they struggle to map these high-level insights into precise pixel-space action. The models excel when reasoning is constricted to discreet, well-scoped dimensions - performance degrades when they are asked to balance abstract aesthetic judgments with deterministic pixel-level corrections. This suggests that the models have trouble bridging symbolic understanding and spatial grounding, especially in iterative workflows that demand surgical image edits.
Sponsored
The world's biggest event for AI builders is in SF from June 3-5. Speakers include leaders from OpenAI, DeepMind, Anthropic, and more, click here to get 30% off your ticket [13].
Google DeepMind's research on defending Gemini against indirect prompt injection attacks reveals that more capable models aren't inherently more secure, and defenses that work in static tests often fail against adaptive attackers. Their findings show that "adversarial training" - fine-tuning models on examples of malicious prompts - significantly improved Gemini 2.5's resistance without degrading performance on normal tasks. When they combined this approach with a "Warning" defense, attack success rates plummeted from 94.6% to just 6.2%.
OpenAI has added built-in tools and features to the Responses API. The API now has support for all remote Model Context Protocol (MCP) servers and tools like image generation, Code Interpreter, and improvements to file search. It has new features that improve reliability, visibility, and privacy for enterprises and developers. All of these tools and features are now available in the Responses API, which supports the GPT‑4o series, GPT‑4.1 series, and o-series reasoning models. Image generation is only supported on o3 in the reasoning model series.
Giving large language models the full output of tool calls is costly and slow. Output schemas enable developers to get structured data, which can then be processed. Using code execution as a way to process data from MCP tools opens the door to scaling the work of AI models. However, allowing an execution environment to also access MCPs, tools, and user data requires careful design as to where API keys are stored and how tools are exposed.
FutureHouse combined specialized literature search agents with a data analysis agent in a continuous experimental loop to accelerate medical discovery. The system autonomously generates hypotheses from literature reviews, proposes experiments for human execution, then analyzes resulting data to inform the next research cycle. Through this process, researchers identified ripasudil, an eye drop medication that relaxes cellular tension, as a promising new treatment for the leading cause of age-related vision loss, where retinal light-sensing cells gradually deteriorate. The complete code, data, and agent interaction traces will be publicly released on May 27.
AI products depend heavily on user prompts, unlike traditional software with predictable outcomes. Skilled prompting can yield varying results due to nuanced intent and context. Product teams can improve results by refining prompts and using follow-up questions to guide users.
v0's AI model, which has specialized web-dev knowledge and an OpenAI-compatible API, is now in beta in the API, AI SDK, and AI Playground.
Gemini Diffusion is Google's first large language model to use diffusion in place of transformers - it has the performance of Gemini 2.0 Flash-Lite at five times the speed.
Meta has announced a new initiative to support early-stage U.S. startups using its Llama models.
LM Arena, a crowdsourced AI benchmarking platform, has secured $100 million in seed funding led by Andreessen Horowitz and UC Investments, despite recent accusations from some researchers that it helps top AI labs game its influential leaderboards.
Extended chain-of-thought reasoning helps large language models better calibrate their confidence.
Sponsored
It doesn't really make sense to run a massive inference model for every small task. That's why Fastino is introducing TLMs [4] - task-specific Language Models that are better, faster, and cheaper.
→ 17% more accurate in task-specific benchmarks
After launching a private beta in December, Google has released Jules to the public. The Gemini 2.5-powered tool clones entire repositories, then autonomously writes tests, fixes bugs, and builds features while developers work elsewhere. The agentic coding landscape is divided between synchronous pair-programming assistants and fully independent agents like Devin and Jules.
Google is reentering the smart glasses space with Android XR, integrating its Gemini AI to deliver real-time vision analysis, translation, and contextual assistance through AR glasses. The rollout will begin with Project Moohan, a mixed-reality headset built with Samsung, and followed by Project Aura, a developer-focused AR glasses prototype from Xreal, and future consumer AI glasses from partners like Warby Parker and Gentle Monster.
Google's Deep Think is an enhanced reasoning mode for the company's flagship Gemini 2.5 Pro model. It can consider multiple answers to questions before responding. Deep Think enabled Gemini 2.5 to top LifeCodeBench, a challenging coding evaluation, and it also beat OpenAI's o3 on MMMU, a test for skills like perception and reasoning. Google will be testing Deep Think with 'trusted testers' and conducting safety evaluations before rolling out the feature widely.
Apple plans to open up its large language models via a software development kit (SDK) so developers can use them in their own apps. The SDK will initially only run Apple's smaller on-device models and won't allow developers to access the models running in the cloud. It will likely be announced during the Worldwide Developers Conference event scheduled to start on June 9. The event will also include an announcement for an overhaul of iOS, macOS, and iPadOS that will make them more unified and resemble the Vision Pro operating system.
A report from Quora's multi-model chat platform Poe revealed dramatic shifts in AI model usage from January to May 2025, with OpenAI's GPT-4.1 family and Google's Gemini 2.5 Pro quickly gaining popularity while Anthropic's Claude models declined. The report highlights clear leaders across categories: general text is dominated by GPT-4.1, reasoning capabilities are led by Gemini 2.5 Pro, image generation is led by Google's Imagen3, and video creation is more split, with Runway currently leading.
KumoRFM is a pre-trained relational foundation model designed to work across any database and predictive task without task-specific training.
Sponsored
The 2025 Semantic Layer Summit brings together experts from Blue Mercury, TELUS, HSBC and Home Depot to explore how semantic layers fuel GenAI, modern BI, and trusted decision-making. Learn how to scale AI with governed data models, reuse KPIs across tools, and accelerate innovation.
Kevin Murphy, a widely-cited Google researcher, just updated his 200-page reinforcement learning textbook, which covers everything from traditional approaches to cutting-edge developments like DPO, GPRO, and reasoning.
François Chollet and the ARC Prize team have released ARC-AGI-2, a significantly more challenging version of their abstract reasoning benchmark. Initial testing shows even the most advanced AI systems struggling, with o3 scoring just 3% compared to 53% on the original benchmark, while human participants successfully solved 75% of tasks on average. The ARC Prize 2025 competition offers $1 million in prizes, including a $700,000 grand prize for the first team to achieve 85% accuracy.
DeepSeek researchers have shared insights from DeepSeek-V3 as a case study in hardware-model co-design. They addressed LLM scaling limitations and using innovations like Multi-head Latent Attention, Mixture of Experts, FP8 training, and a Multi-Plane Network Topology to maximize GPU efficiency and minimize communication overhead.
Google Meet now supports real-time speech translation, using DeepMind's audio language model to preserve voice, tone, and expression across languages.