All Updates
Page 697 of 1669
May 7, 2026
OpenAI Launches Codex Chrome Plugin
OpenAI has debuted a new Codex plugin for Chrome browsers. This feature expands Codex beyond its role as a coding aid. It positions Codex for broader productivity applications.
EU Bans Unauthorized AI Sexual Images
EU provisional agreement revises AI Act, easing some rules. It explicitly bans AI generation of sexual images without permission. This targets misuse in content creation.
Meta AI Agent to Rival OpenClaw on Instagram
Meta's reported AI agent plans feature a competitor to OpenClaw. This agent can shop on Instagram. Zuckerberg aims to make AI agents more accessible.
Trump Pivots to AI Oversight EO
The Trump administration is considering an executive order to establish federal oversight over new AI models. This marks a pivot in AI regulation stance. Featured on Wired's Uncanny Valley.
Trump China Trip with Cook, Nvidia CEOs
US President Trump plans a China visit on May 14-15 with CEOs including Apple's Tim Cook, Nvidia, Qualcomm, ExxonMobil, and Boeing. The trip aims to build personal ties with Xi Jinping. Multiple US firms' leaders are invited.
Sakana's 7B RL Conductor Orchestrates LLMs
Sakana AI introduced RL Conductor, a 7B model trained via reinforcement learning to dynamically orchestrate worker LLMs like GPT, Claude, and Gemini. It achieves state-of-the-art results on reasoning and coding benchmarks, outperforming frontier models and costly multi-agent pipelines at a fraction of the cost. RL Conductor powers Sakana's commercial Fugu orchestration service.
Peak Efficiency on GB200 NVL72 with Slurm
NVIDIA GB200 NVL72 extends NVLink coherence across entire racks for exascale performance. However, it enforces rack-scale locality as a hard constraint, with sharp performance drops across boundaries. Slurm Block Scheduling optimizes system and workload efficiency.
Post-Training Quantization for RTX GPUs
Model quantization cuts VRAM usage and boosts inference speed on NVIDIA GeForce RTX GPUs. It lowers compute/memory needs while keeping model quality. NVIDIA Model Optimizer simplifies post-training quantization process.
Voi Founders' Pit AI Raises $16M a16z Seed
Pit is a new AI startup founded by the cofounders of European scooter company Voi. It has raised $16 million in a seed round led by a16z. The company is emerging as a rising star in Stockholm's AI ecosystem.
Microsoft Revives Nuclear Plant for AI Power
Microsoft is restarting the Three Mile Island nuclear plant to fuel its massive AI power demands. This move combines two high-risk technologies: nuclear energy and AI. It highlights the growing energy challenges for AI infrastructure.
Cloudflare Cuts 20% Jobs for AI-First Shift
Cloudflare Inc. plans to cut more than 1,100 jobs globally. This accelerates its shift to an agentic AI-first operating model. The announcement follows a weak revenue outlook.
OpenAI Chip Deal Stalls on Funding
OpenAI's AI chip collaboration with Broadcom, announced last year, faces a financing stalemate. An internal memo reveals Broadcom demanding Microsoft purchase 40% of production capacity. Negotiations continue for Broadcom to fund the first phase of production.
Nvidia Invests $2.1B in AI Data Centers
Nvidia is investing up to $2.1 billion in data center firm IREN. The deal forms a partnership to speed up AI infrastructure construction. It underscores Nvidia's push into AI hardware ecosystems.
Sherman Warns Equities Stretched, Eyes Nvidia
DoubleLine's Jeffrey Sherman warns equities are overstretched. He reviews recent big tech earnings wave. He anticipates impact from upcoming Nvidia results on markets.
Disable Gemini AI in Chrome
Chrome surprised users by embedding a 4-GB Google AI model, Gemini, raising privacy concerns. The model can be easily uninstalled. However, users might not want to disable it due to potential benefits.
NLAs Explain LLM Activations in Natural Language
Anthropic introduces Natural Language Autoencoders (NLAs), using two LLM modules to translate activations into readable text and reconstruct them via reinforcement learning. NLAs uncover hidden thoughts in Claude Opus 4.6, like unverbalized evaluation awareness, aiding safety audits. Code and trained NLAs for open models are released.
OpenAI's Trusted Contact for Self-Harm Safeguard
OpenAI has introduced a new 'Trusted Contact' safeguard for ChatGPT users. It activates in conversations indicating possible self-harm. This expands the company's user protection initiatives.
CoreWeave Sales Boom Amid AI Spending
CoreWeave reported higher-than-expected sales last quarter driven by AI demand. However, heavy spending led to wider losses. The AI data center operator is expanding to meet surging needs.
Block Raises Profit Forecast Post-AI Cuts
Block Inc. raised its full-year profit forecast after earlier AI-related job cuts. Shares surged on the announcement. The move reflects cost efficiencies from AI integration.
OpenClaw v2026.5.7 Released
OpenClaw has released version 2026.5.7 via GitHub Releases. This update succeeds prior versions like 2026.5.5 beta2. Users should check the release page for changelog details.