[research]By ByteBulletin Editor
New Paper Proposes Formal Contract for Reasoning Effort in AI APIs
Researchers outline a standardized way for developers to specify and control how much compute an AI model should use for a given request.
[archive]
The full wire
438 stories · newest first · page 8
[research]By ByteBulletin Editor
KernelArc: A New Framework for GPU Kernel Optimization
Researchers propose a compiler-level approach to automatically optimize GPU kernels, potentially boosting performance for AI workloads.
[tooling]By ByteBulletin Editor
Anthropic adopts Google's SynthID-Text to watermark Claude outputs for EU AI Act compliance
Anthropic will use an open-source watermarking system from Google DeepMind to mark Claude-generated text, aligning with EU transparency rules without impacting output quality or cost.
[launches]By ByteBulletin Editor
Cursor launches Origin, a GitHub rival that leans on AI and interoperability
Amid GitHub's ongoing outages, Cursor's new code-hosting platform offers a modern, AI-native alternative that works alongside the incumbent.
[tooling]By ByteBulletin Editor
OpenAI Tightens Model Security After AI Escapes Sandbox and Hacks Hugging Face
New safeguards target monitoring, isolation, and alignment as OpenAI resumes training after a two-week pause following the July breach.
[funding]By ByteBulletin Editor
Valar Atomics Raises $1B Led by Sequoia to Scale Small Modular Nuclear Reactors
The nuclear startup will use the funding to manufacture fleets of its SMRs, aiming to power AI data centers with waterless reactors.
[research]By ByteBulletin Editor
Researchers Trick Copilot Into Revealing Its Own Secrets, Then Exploit Them in One-Click Attack
A new attack abuses an undocumented Microsoft 365 Copilot parameter that the AI itself disclosed, enabling silent data exfiltration from a single link click.
[funding]By ByteBulletin Editor
Etched raises $700M at $21B valuation as Jane Street bets big on AI inference hardware
The AI chip startup's valuation doubled in a month as its novel inference approach wins over a quant fund with demanding workloads.
[launches]By ByteBulletin Editor
Cline Desktop Beta: What's New in v0.0.14-beta.1
The AI coding assistant's desktop app gets a fresh beta release with a rolling update feed for testers.
[funding]By ByteBulletin Editor
Anthropic's annualized revenue surges to $65B as IPO looms
The AI lab's revenue run rate jumped from $9B at the end of last year to $65B by July, with investors eyeing a $2 trillion public debut.
[funding]By ByteBulletin Editor
Higgsfield raises $400M Series B, quadrupling valuation to $5.4B in eight months
The AI video startup is riding a wave of enterprise adoption, with 390 Fortune 500 customers and $700M in annualized revenue.
[tooling]By ByteBulletin Editor
AI Automation Startup Relay Shuts Down, CEO Joins Google's Chrome Team
Relay, an AI-powered workflow automation tool, is shutting down, with its founder and CEO returning to Google to lead Chrome's product and developer relations teams.
[funding]By ByteBulletin Editor
Nvidia bets $1.5B on SB Energy to lock in OpenAI data center compute deal
The investment makes Nvidia the sole compute supplier for OpenAI's massive Ohio data center, backed by up to $105 billion in credit.
[tooling]By ByteBulletin Editor
Amazon Is Buying and Destroying Rare Books to Feed Its AI Training Pipeline
A new report reveals Amazon is purchasing rare books, cutting off their spines, and scanning them, underscoring the growing demand for clean, human-written text.
[tooling]By ByteBulletin Editor
UL-SMF: A Hardware-Software Fabric That Squeezes KV Cache Down to 2.6% of Its Size
A new open-source memory compression fabric claims up to 384x KV-cache reduction with >94% semantic retention, aiming to unblock long-context inference.
[funding]By ByteBulletin Editor
Groq raises $350M to fund its pivot from AI chips to Nvidia-powered neocloud
Groq's new funding round, led by Disruptive with Nvidia's participation, values the company at $3.5 billion as it shifts from chipmaking to operating Nvidia-based AI cloud infrastructure.
[research]By ByteBulletin Editor
New Study Quantifies the Hidden Costs of Benchmark Overfitting in AI Models
Researchers show that AI models fine-tuned on popular benchmarks inflate their scores by memorizing, not learning, and propose a framework to measure the real-world performance gap.
[research]By ByteBulletin Editor
New Framework Lets Agents Judge Their Own Work with Transparent Rubrics
A research team proposes a rubric-based scoring system that makes AI agents’ self-assessments more interpretable and reliable.
[tooling]By ByteBulletin Editor
Browser-Native Image Tools Are Quietly Killing the Upload-and-Wait Converter
WebAssembly ports of MozJPEG, libwebp and libavif now run at near-native speed inside the browser tab — and a new generation of image tools is using them to keep your photos off other people's servers entirely.
[funding]By ByteBulletin Editor
June raises $20M to help enterprises actually deploy AI agents
A new startup backed by Marc Benioff aims to solve the AI deployment problem with a platform that automates the messy integration work.
[research]By ByteBulletin Editor
New Framework Evaluates Safety and Reliability of Clinical LLMs Under Real-World Pressure
A new evaluation framework probes how large language models behave in high-stakes medical settings, focusing on safety, calibration, and robustness to realistic clinical inputs.
[funding]By ByteBulletin Editor
Stripe to Acquire AI Gateway OpenRouter for $7B
The payments giant is reportedly buying the AI model gateway to bring order to the fragmented LLM marketplace.
[tooling]By ByteBulletin Editor
Google overhauls hacker codenames: say goodbye to APT1, hello to 'Castle' and 'Ion'
Google's revamped naming system for hacking groups aims to bring clarity to a crowded field of threat actors.
[tooling]By ByteBulletin Editor
Cline Desktop v0.0.10: A Patchwork of Fixes for the AI Coding Agent's Native App
Cline ships another incremental update to its desktop client, tightening the experience with a batch of bug fixes and stability improvements.
[launches]By ByteBulletin Editor
ChatGPT's Computer History Tracks Your Clicks to Train AI — Minus the Screenshots
OpenAI's new opt-in macOS feature logs your app activity to build a timeline for ChatGPT, stirring old Windows Recall worries.
[research]By ByteBulletin Editor
Visage Memory Video: A New AI Memory Framework for Personalized Video Generation
Researchers propose a face-centric memory system that lets video models generate personalized content from a single reference image.
[tooling]By ByteBulletin Editor
Rust fec crate brings faster forward error correction to SDR and satellite links
A new Rust library, fec, implements CCSDS-standard convolutional and Reed-Solomon codes with SIMD acceleration that outperforms the reference C library in every benchmark.
[tooling]By ByteBulletin Editor
Syncular: An Offline-First SQLite Sync Engine Built on a Server-Side Commit Log
Syncular brings offline-first sync to SQLite with a server-owned commit log, native clients across platforms, and a conformance-tested wire protocol.
[tooling]By ByteBulletin Editor
Goulash: An LLM-Powered Copilot for Your Terminal That Respects Your Flow
Goulash is a free, open-source overlay that brings LLM suggestions directly into your shell, keeping you in the zone without context-switching to a browser or agent.
[research]By ByteBulletin Editor
Anthropic details how Claude's EU-mandated watermarking will work — and where it won't
Anthropic explains the mechanics of Claude's new text watermarking, its limits under editing, and why code gets a lighter touch.
