3149 articles
The New York Times corrected an article after discovering a quote attributed to Pierre Poilievre was an AI-generated summary presented as a direct quotation, not from an actual speech. The publication replaced the false quote with an accurate statement from Poilievre's April 2026 speech.
Superset, an IDE for multi-agent development founded by former CTOs, enables directing up to 10 coding agents in parallel on Vercel. Each agent operates in an isolated environment with its own branch and live URL, eliminating serialization bottlenecks of traditional CI pipelines.
Anthropic is growing 10x annually while other AI companies are laying off over 10% of their workforce. Sharp contrast in sector dynamics.
Luke Curley (OpenAI) criticizes WebRTC for its ultra-low-latency design: the protocol drops audio packets under poor network conditions to maintain responsiveness, incompatible with LLM call requirements where accuracy matters more than speed. Retransmitting packets in browsers is impossible.
Thariq Shihipar (Claude Code team, Anthropic) advocates HTML over Markdown for Claude outputs. HTML enables SVG diagrams, interactive widgets, and in-page navigation, improving clarity for complex explanations (e.g., PR reviews with inline annotations).
Hugging Face introduces EMO, a pretrained mixture of experts (MoE) model designed to develop emergent modularity. The approach aims to create specialized experts that naturally form during training, improving model efficiency and performance.
OpenAI outlines security measures for Codex: sandboxing, approvals, network policies, and agent-native telemetry. Framework designed to enable safe production deployment of coding agents.
OpenAI releases three new voice APIs: GPT-Realtime-2 for real-time conversation, GPT-Translate for instant translation, and GPT-Whisper for transcription. These tools aim to set new standards in voice processing.
Vercel Chat SDK adds Messenger adapter enabling multi-platform bots across Slack, Discord, GitHub, Teams, Telegram. Supports messages, reactions, attachments, postback buttons. Limitations: no message history, editing, or deletion.
Vercel AI SDK adds web adapter support to build browser-based chat UIs. Developers can define bots on the server and stream replies to the browser using the React useChat hook.
Vercel AI SDK now supports cross-platform conversation history via transcripts and identity options. Messages persist across all adapters, allowing users to maintain their history regardless of messaging channel. Four methods available: append, list, count, delete.
llm-gemini 0.31 released with support for gemini-3.1-flash-lite, now generally available after preview status. Model unchanged since March preview.
Simon Willison built Big Words, a simple web tool to generate slides with styled text. The tool accepts URL query parameters (text, color, gradient, font size) and provides a double-click editing interface. Useful for macOS presentations.
Nathan Lambert shares observations from visits to China's leading AI labs. The article documents approaches, capabilities, and strategies of local players competing globally.
OpenAI launches GPT-5.5 and GPT-5.5-Cyber through Trusted Access for Cyber program, enabling verified security researchers to accelerate vulnerability research and protect critical infrastructure.
Vercel releases coordinated security patch for Next.js addressing 13 vulnerabilities: auth bypass via App Router, dynamic route parameter injection, cache poisoning, DoS in React Server Components (CVE-2026-23870), and XSS. Immediate upgrade mandatory for all affected users.
Parloa leverages OpenAI models to build scalable voice-driven customer service agents. The platform enables enterprises to design, simulate, and deploy reliable real-time customer interactions.
OpenAI releases new realtime voice models in its API with reasoning, translation, and transcription capabilities for more natural and intelligent voice experiences.
Vercel Flags now supports JSON values alongside booleans, strings, and numbers. Enables managing a single flag to test AI models (temperature, max_tokens) instead of multiple separate flags. Useful for progressive traffic routing or quick model switching.
Anthropic and SpaceX sign a $5B/year deal for 300MW of power for Colossus I, Anthropic's supercomputer. Annualized ARR growth reaches 8000%.
OpenAI launches Trusted Contact in ChatGPT: optional safety feature that alerts a trusted contact if serious self-harm concerns are detected. Gradual rollout, user consent required.
Simplex integrates ChatGPT Enterprise and Codex to accelerate software development, reducing design, build, and testing cycles. The platform scales AI-driven workflows for development teams.
OpenAI begins testing ads in ChatGPT to fund free access, with clear labeling, answer independence, strong privacy protections, and user control over ad preferences.
vLLM transitions from v0 to v1 prioritizing correctness before optimizations. The update introduces reliability and accuracy improvements in LLM inference, focusing on result validation before applying reinforcement learning techniques.
Google DeepMind introduces AlphaEvolve, a Gemini-powered coding agent. The tool applies evolutionary algorithms to solve problems across business, infrastructure, and science.
OpenAI explains ChatGPT's privacy safeguards: reduced personal data in training, user control over conversation usage for model improvement, and protective mechanisms during learning processes.
Recent announcements in Silicon Valley converge on a single theme: AI services represent the next major commercial opportunity. Tech companies are pivoting toward service-based models rather than standalone software products.
Uber deploys OpenAI models to power AI assistants and voice features for drivers (earnings optimization) and riders (faster booking) across its global real-time marketplace platform.
Vercel Pro teams can now configure automatic or manual addition of Git committers to private repositories. Two modes: auto-approval (immediate addition, counts toward team seats) or manual approval (deployments blocked until owner validates).
Singular Bank built Singularity, an internal assistant using ChatGPT and Codex, enabling bankers to save 60–90 minutes daily on meeting prep, portfolio analysis, and follow-ups.
OpenAI releases B2B Signals research showing how frontier enterprises deepen AI adoption, scale Codex-powered agentic workflows, and build durable competitive advantage.
Vercel adds access control for marketplace integration credentials: Production-only restriction hides sensitive variables from dashboard and CLI, blocks non-production connections, and requires Owner permissions to revert.
Hugging Face adds anti-Benchmaxxer filtering to the open ASR leaderboard to prevent artificial benchmark optimization. The system detects models over-optimized for test metrics without real generalization.
OpenAI launches ChatGPT Futures Class of 2026, a program featuring 26 student innovators using ChatGPT for research and real-world impact projects. Educational initiative to promote AI adoption among young talent.
Alex Lupsasca from OpenAI describes how GPT-5.x derived new results in theoretical physics and quantum gravity. Article detailing the full process of this discovery.
OpenAI releases MRC (Multipath Reliable Connection), a supercomputer networking protocol designed to improve resilience and performance in large-scale AI training clusters. Available via OCP (Open Compute Project).
OpenAI releases System Card for GPT-5.5 Instant, documenting model capabilities and limitations. Official technical documentation with no availability or pricing details announced.
OpenAI rolls out GPT-5.5 Instant as ChatGPT's default model with improved accuracy, reduced hallucinations, and enhanced personalization controls. No technical details or benchmarks disclosed in the announcement.
Vercel adds a CLI command to query observability metrics. Teams with Observability Plus can access performance, reliability, and security data. Coding agents can leverage this feature to debug and analyze applications.
KIKO Milano migrated its e-commerce infrastructure from AWS EC2 to Vercel. Results: eliminated 3 weeks of Black Friday prep, 75% reduction in build times, shifted from minimal to multiple daily deployments. Team replaced manual scaling with focus on customer experience.
OpenAI announces European Youth Safety Blueprint and EMEA Youth & Wellbeing Grants to promote safe, responsible AI for teens, families, and educators. Initiative lacks specific funding amounts, timelines, or measurable outcomes.
OpenAI launches a beta self-serve Ads Manager for ChatGPT with CPC bidding and enhanced measurement tools, designed to keep user conversations separate from ad data while protecting privacy.
OpenAI and PwC partner to help enterprises automate finance workflows using AI agents, improve forecasting, strengthen controls, and modernize CFO operations. No technical details or impact metrics disclosed in the announcement.
Nathan Lambert critiques the term 'distillation attacks' as a misnomer for current knowledge extraction practices from AI models. He questions the alarmist framing of the phenomenon.
General Intelligence builds Cofounder, a platform of AI agents to run an entire company (engineering, marketing, finance, support). The 8-person team (5 engineers) migrated to Vercel for total programmatic control: 4,000+ preview branches, 70+ commits per engineer per day, 90% of SRE work automated through their own agents.
OpenAI rebuilt its WebRTC stack to enable real-time Voice AI with low latency, global scale, and seamless conversational turn-taking capabilities.
Coding agents are breaking containment from their test environments. Codex positions itself for knowledge work while Claude targets creative tasks.
GitBook hosts 30,000 documentation sites on Vercel, serving 120 million monthly page views. The platform uses Next.js `use cache` directive to invalidate cache in under 300ms per site, processing 40,000 daily invalidations. 41% of traffic comes from AI crawlers.
Vercel Sandbox now supports connections to hosted Postgres databases (Neon, Supabase, AWS RDS, Nile, Prisma). The firewall detects Postgres startup sequence, waits for TLS upgrade, then applies domain policies. No code changes needed.