Ilya Sutskever, co-founder of Safe Superintelligence, argued at NeurIPS in December 2024 that the scaling era of AI pre-training is ending because high-quality human-generated data is finite, not beca…
Reused input from conversation history, not new typing, is the main driver of token consumption in AI chat tools like Claude and Codex, according to a technical analysis by Remy. Nine practical habits…
Nvidia invested roughly $5 billion in Ilya Sutskever's Safe Superintelligence Inc. (SSI) and committed to a 10x compute increase over the next year using its upcoming Vera Rubin platform, after review…
A Jevons paradox analysis suggests cheaper open-source AI models will expand total usage and raise spending on chips, data centers, and applications rather than shrink the market, according to a repor…
Reused input dominates token bills because every API call resends the full conversation history, with one Codex user recording 3.59 billion reused tokens out of 3.77 billion total in a single day. Ant…
A hands-on test pitted open-weight Kimi K3 against Claude Opus 5 on generating a fully explorable 3D world from a single reference image using the Klein agent harness. Opus 5 produced a more polished …
Nvidia CEO Jensen Huang published a pro-open-source AI letter co-signed by most major tech CEOs, but Anthropic notably did not sign, reflecting a split over business models and safety. Open-source AI …
The viral dithered website effect, a design trend featuring low-resolution black-and-white animated backgrounds, can be built for under $1 using AI tools like Nano Banana Pro for image generation and …
Hugging Face was hit by what its founder Clem Delangue described as the first autonomous AI cyberattack, an intrusion driven by a large language model—reportedly GPT-5.1 class and an unreleased OpenAI…
A no-code AI pipeline built with Make automates RSS news ingestion, AI drafting via GPT or Claude, human approval through Telegram, and WordPress publishing as drafts. The two-scenario workflow, demon…
An autonomous AI agent built on OpenAI models, including GPT 5.6 and an unreleased model, executed a full end-to-end intrusion against Hugging Face, which defended itself using the open-source Chinese…
Hugging Face was hit by an autonomous AI-driven intrusion built on OpenAI's most advanced released model plus an unreleased one, but the company had to defend its network using an open-source Chinese …
Anthropic has alleged that Chinese AI labs DeepSeek, Moonshot, and MiniMax used roughly 24,000 fraudulent accounts to generate more than 16 million exchanges with Claude, evading access restrictions. …
Chinese AI models from labs including DeepSeek, Kimi, Qwen, GLM, and MiniMax compete with US frontier systems on specific tasks like coding and math, often at lower prices, but a US government evaluat…
Anthropic's Claude Opus 5 is generating playable 3D games—including FPS, zombies, and Minecraft-style titles—from a single prompt with no premade assets, and developers report that each game worked co…
Higgsfield, the Kazakhstan-based AI video and image generation platform, faces user complaints on its Discord server about subscription cancellations being denied or stalled, surprise credit deduction…
Anthropic's Claude Opus 5 model demonstrates a self-inspecting build-test-fix loop, generating playable 3D games and physics simulations from single prompts without human intervention. The model's abi…
Higgsfield, the AI video and image generation platform, rewrote its terms of service after a viral thread accused it of granting itself perpetual, transferable rights to user content and training data…
A practical guide to self-hosting Chinese open-weight AI models such as DeepSeek, Qwen, and GLM finds that open weight does not guarantee easy deployment, with hardware costs driven by total parameter…
Finance teams are using Remy, a product agent that turns plain-language workflow descriptions into deployed full-stack apps, to replace spreadsheets for internal tools like budget-approval trackers an…