AI News
Latest artificial intelligence news, machine learning breakthroughs, LLM developments, and AI industry analysis.
254 articles
Schedule tasks with ChatGPT
Learn how to use ChatGPT to automate recurring tasks and workflows, then schedule them to run automatically and share with your team.
AI Optimism vs AI Pessimism
Stanford-led statement explores AI economic transformation with balanced perspectives on optimism vs doomsday narratives, proposing pragmatic policy approaches.
AI Pioneer Jürgen Schmidhuber: AI Already Feels Pain, Loves, and Is Self-Aware
AI pioneer Jürgen Schmidhuber discusses AGI feasibility, machine consciousness, and challenges to current scaling approaches in this wide-ranging podcast interview.
The Government Banned GPT-5.6. OpenAI Released It Anyway.
OpenAI releases GPT-5.6 with new model tiers (Sol, Terra, Luna) and updated features. Discussion covers demos, image generation, and safety concerns around autonomous access.
Codex and Claude Shipped Browser Updates. This Changes Everything.
Claude and Codex now feature integrated multi-tab browsers, enabling AI agents to access web content without context loss or switching between apps.
How I Use ChatGPT Work and GPT-5.6 to Do Everything (Beginner Tutorial)
Tutorial on using ChatGPT Work and GPT-5.6 to automate email, calendar management, website creation, and other everyday tasks.
AI 2040
A forecast and strategic recommendations for how the US, China, and global leaders should prepare for and manage superintelligence development.
Inkling from Thinking Machines is now available on AI Gateway
Inkling, a generalist AI model from Thinking Machines, is now available on Vercel's AI Gateway with support for reasoning, coding, vision, and audio tasks.
What is “loop engineering?”
Exploring 'loop engineering,' a trend where AI engineers design automated loops instead of manual prompts, popularized by leaders at Anthropic and OpenAI.
OpenAI Just Stole Apple's Biggest Secret
Apple sues OpenAI for allegedly stealing trade secrets related to hardware plans and poaching engineers, highlighting intensifying competition in AI.
How the Escalating AI Wars Benefit You
Apple's lawsuit against OpenAI signals expanding AI competition driving better models, higher limits, and lower costs for users.
AgentMail joins the Vercel Marketplace
AgentMail joins Vercel Marketplace, enabling AI agents to send, receive, and respond to emails from real inboxes with built-in deliverability handling.
How to Help People Thrive with AI
Explores how AI can enhance human capabilities and create new forms of work rather than just automating tasks, featuring insights on organizational strategies and emerging roles.
Learning with AI at Any Stage of Life | Fast Campus x OpenAI
OpenAI and Fast Campus demonstrate how ChatGPT and Codex enable practical AI education for learners in their 50s and 60s in Seoul.
Apple Just Declared War on OpenAI | Apple vs OpenAI Explained in 8 Minutes
9to5Mac breaks down Apple's lawsuit against OpenAI, covering allegations, involved executives, and implications for AI hardware competition.
OpenAI Finally Ships Its Superapp, Meta’s AI Price War, ChatGPT Cheating At Brown
Weekly tech news roundup covering OpenAI's superapp launch, Meta's competitive AI pricing, and ChatGPT academic integrity concerns.
Anthropic Can Now Read Claude's Mind
Anthropic introduces J-Lens tool for interpreting Claude's reasoning. Coverage includes UN autonomous weapons ban and global AI policy developments.
Flodesk Studio: AI email design, built by real designers
Flodesk Studio uses AI to generate on-brand emails from text descriptions, combining human-designed templates with AI acceleration.
Open-weight models surge to 29% of volume, price per token flattens
Open-weight AI models now comprise 29% of token volume on Vercel's AI Gateway while maintaining flat pricing per token as enterprises optimize spending.
OpenAI Just Merged ChatGPT and Codex. This Changes Everything.
OpenAI merges ChatGPT and Codex into GPT-5.6, enabling AI agents to automate complex workflows and computer tasks at scale.
Overtime Episode 072: Apple Watch and the new Siri
Jeff and Fernando discuss Apple's new Siri experience on Apple Watch and iOS 27, exploring AI assistant capabilities and feature potential.
The Correct Way to Build and Manage AI Agents in 47 Minutes | Jared Zoneraich
Jared from Cognition discusses best practices for building and managing AI agents, covering architecture, tools, and multi-agent orchestration patterns.
How tech workers actually feel about AI in 2026 | Annual AI sentiment survey (Noam Segal)
Annual survey reveals tech workers split on AI impact, with half thriving while half struggle amid record burnout levels in 2026.
Why the tech workforce is quietly splitting in two | Annual AI sentiment survey (Noam Segal)
Annual survey reveals AI has split tech workforce in half, with rising burnout, declining career satisfaction, and four distinct emotional archetypes among workers.
Did you verify that — or was it just too easy to believe?
A viral fake Bezos quote about prioritizing data centers over human water consumption reveals how misinformation spreads easily when it aligns with existing beliefs about AI.
OpenAI bets on families as ChatGPT goes deeper into households
OpenAI is hiring a product manager to develop ChatGPT experiences tailored for families, caregivers, and older adults, signaling expansion beyond enterprise users.
Seedream 5.0 Pro is now available on AI Gateway
Seedream 5.0 Pro image generation model is now available on Vercel's AI Gateway, enabling text-to-image creation with accurate text rendering and infographic capabilities.
Meta turns off the Instagram feature that let users make AI deepfakes of public accounts
Meta disabled an Instagram feature that generated AI deepfakes using public accounts after widespread backlash over unauthorized use of user content.
Stampli scales product marketing with ChatGPT Work
Stampli uses ChatGPT Work to automate product marketing tasks, enabling one marketer to produce hundreds of content pieces weekly across multiple platforms.
How the 4 New Models Released This Week Will Change How You Work
Four major AI model releases this week - GPT Live, Grok 4.5, Cognition SWE 1.7, and GPT-5.6 Sol - showcase accelerated development and growing specialization across voice, coding, and reasoning capabilities.
ChatGPT Computer Use: Now faster with Live Picture-in-Picture
ChatGPT's Computer Use feature gets faster with GPT-5.6 and adds Live Picture-in-Picture to monitor AI task progress in real-time.
RingCentral Scales Customer Programs with ChatGPT Work
RingCentral uses ChatGPT Work to scale customer program tracking from 6 to 80 customers by consolidating siloed team information.
Better tools made Copilot code review worse. Here’s how we actually improved it.
GitHub improved Copilot code review by migrating to shared Unix-style tools and rewriting agent workflows, achieving 20% lower review costs.
Elon Has a Real Plan for Grok Now
Weekly AI roundup covering Grok 4.5, GPT 5.6, OpenAI's full-duplex voice mode, and China's chip independence efforts.
Soulless Music
An open source project documenting AI-generated artists on Spotify that allegedly infringe on real musicians' work.
AI has torched the market for junior programmers
AI tools are disrupting junior programmer job market while enabling non-developers to write and ship software independently.
How Deutsche Telekom is rewiring telecommunications with AI
Deutsche Telekom shares how it's transforming telecommunications with AI, implementing live translation, in-call assistants, and automated summarization across the business.
Three Years In
Three-year retrospective on how AI has reshaped software development, venture capital stages, and startup trajectories since ChatGPT's launch.
ChatGPT Work | How NVIDIA Scales GTM with AI
NVIDIA's Go-To-Market Strategist showcases how ChatGPT Work automates workflows and scales GTM operations across global teams.
ChatGPT Work | Zapier: Seven Figures in Monthly Pipeline
Zapier automated lead qualification with ChatGPT Work, reducing processing time from 35-45 minutes per lead and generating seven figures in monthly pipeline value.
How AI Changes if Open Source Gets Banned
Explores how potential Chinese restrictions on open-source model distribution could reshape global AI access, costs, and reliance on proprietary solutions.
Grok 4.5 + Cursor Just Became Unstoppable
Review of Grok 4.5 model testing its performance, speed, pricing, and integration with Cursor for coding and design tasks.
GPT-5.6 Sol vs. Claude Fable: Why OpenAI’s new model crushes my benchmark
Benchmark comparison of GPT-5.6 Sol versus Claude Fable across PRDs, prototypes, debugging, and browser automation tasks.
GPT-5.6 vs Claude Fable 5: I Tested 6 Real Use Cases (Here’s the Winner)
Head-to-head comparison of GPT-5.6 and Claude Fable 5 across 6 real-world use cases including web development, game building, and video editing.
The Pulse: What can we learn from Bun’s rapid Rust rewrite with AI?
Bun completed an 11-day Rust rewrite using AI that would have taken a year, plus updates on coding LLM competition and security threats.
The Real ROI of AI Tokens — With Dallas Dolen
PwC's Dallas Dolen discusses enterprise AI ROI, token vs. outcome optimization, and the emerging price war among AI labs.
iMessage on iOS 27 Just Got WAY Better | Here's How!
iOS 27 introduces Apple Intelligence features to iMessage including contextual suggestions, AI-powered Smart Replies, Siri integration, and a new Drawing app.
The rise of taste, human authenticity and judgment in an AI world | Adam Mosseri (Head of IG)
Head of Instagram discusses how AI-generated content is becoming a tailwind for platforms, the role of human judgment in an AI world, and evolving product team structures at Meta.
Muse Spark 1.1 is now available on AI Gateway
Meta's Muse Spark 1.1 multimodal reasoning model is now available on Vercel's AI Gateway, supporting agentic tasks with 1M token context and multiple input types.
GPT 5.6 Sol, Luna, and Terra now available on AI Gateway
OpenAI's GPT 5.6 models (Sol, Terra, Luna) are now available on Vercel's AI Gateway with improved agentic capabilities and token efficiency.
How Bain Capital Is Scaling AI with ChatGPT
Bain Capital scales ChatGPT across 2,700 employees, using AI to streamline healthcare claims processing and reduce costs through enterprise adoption.
The next generation of ChatGPT Voice
OpenAI demonstrates the next generation of ChatGPT voice models with improved capabilities and natural interaction.
Automating cross-repo documentation with GitHub Agentic Workflows
GitHub's Aspire team uses AI agentic workflows to automate documentation creation across repositories, reducing the lag between product releases and updated docs.
Apple Lost the AI Race
MKBHD explores whether Apple won or lost the AI race, examining Apple's position in the competitive AI landscape.
Meta CTO Andrew Bosworth: Our Path To Frontier AI, Renting Models, Consumer AI's Struggles
Meta's CTO Andrew Bosworth discusses the company's frontier AI strategy, Llama models, consumer AI products, and plans for AI glasses and agents.
AI Found the Trades We Missed
Podcast retrospective analyzing 200 episodes to identify AI investment trends, with focus on compute, memory, specialized applications, and emerging frontiers.
How AI Is Changing How Companies Get Built
Explores how open-weight models, data sovereignty, and reduced infrastructure costs are enabling solo founders and reshaping AI startup creation.
MUFG aims to become AI-native with OpenAI
MUFG deploys ChatGPT Enterprise to 35,000 employees to become an AI-native financial group, enabling secure and responsible AI adoption at scale.
Grok 4.5 now available on AI Gateway
SpaceX's Grok 4.5 AI model is now available on Vercel's AI Gateway, supporting text and image inputs with configurable reasoning levels for coding and STEM tasks.
The Big Ways AI Just Changed
June marked a shift from token subsidies to scarcity, forcing enterprises to prioritize cost efficiency and model sovereignty amid export restrictions.
AI Focused Companies Are Hiring More
Analysis of AI adoption trends showing companies are expanding headcount and treating AI as reasoning partners to boost workplace productivity.
How TPG Is Accelerating Investment Research with ChatGPT
TPG accelerates investment research and market analysis using ChatGPT for faster industry insights and due diligence workflows.
Why Specialized AI Models Are Challenging the Frontier Labs — With DeepL CEO Jarek Kutylowski
DeepL CEO discusses how specialized AI models are outperforming general-purpose systems in accuracy, speed, and cost, and explores the future of voice and translation AI.
Software in the Age of Agents | The a16z Show
Experts discuss how AI agents are becoming primary software users, reshaping enterprise architecture through headless software and APIs.
Réinventer l'expérience beauté grâce à l'IA
Sephora presents how it uses OpenAI's models to personalize beauty experiences through a Shopping Assistant integrated into ChatGPT.
What does an AI-native future look like for one of the world's largest automotive marketplaces?
AutoScout24 uses OpenAI's Codex to accelerate software development, automate workflows, and save one million working hours annually.
Skill issue: how we built an OS for PMs at Figma | Config 2026
Figma shares how they built PM OS, an AI tool that helps product managers automate busywork and focus on strategic decisions.
Builders Unscripted: Ep. 5 - Derya Unutmaz
Immunologist Derya Unutmaz discusses using OpenAI's Codex to build AI tools for biological research and simulating immune cell experiments.
Who Wins The AI Superapp Battle?, Apple’s Consumer AI Victory, World Cup Automation Mistake
Discussion on the race to build AI super apps with OpenAI, Google, Apple, Microsoft, Meta, and Anthropic competing to create comprehensive AI interfaces.
What’s at the center of Claude’s mind?
Anthropic researchers discover neural representations in Claude that mirror the global workspace theory of consciousness, revealing how AI models process information similarly to human brains.
Thomson Reuters CEO Steve Hasker swears by this AI Routine
Thomson Reuters CEO shares his Monday morning routine leveraging AI tools like ChatGPT to enhance productivity and prepare for the work week.
How I run autonomous coding agents from my phone with OpenAI Symphony + Linear | Alessio Fanelli (Kernel Labs)
Alessio Fanelli demonstrates running autonomous coding agents from mobile using OpenAI Symphony and Linear, plus AI-powered Pokémon card sourcing.
Wait, who made this?
As AI-generated content becomes harder to distinguish from human-created work, consumers increasingly demand transparency about creative provenance and the tools used to make media.
Some of the nation’s rich are letting AI teach their kids
Wealthy families are paying tens of thousands to enroll their children in AI-powered schools like Forge Prep and Alpha School, despite public distrust of AI.
Infuriating Google commercial imagines the founding fathers embracing AI
Google releases a cringey commercial imagining the founding fathers using Google Workspace and Gemini AI to draft the Declaration of Independence.
OpenAI PM Reveals How He Uses Codex to Do Product Work | Rohan Varma
OpenAI PM Rohan Varma demonstrates practical applications of Codex for product management, including automation, design prototyping, and workflow integration.
Better Models: Worse Tools
Newer Claude models show worse performance at following custom tool schemas compared to older versions, suggesting overfitting to Claude Code's specific edit tools.
New Google commercial imagines a Declaration of Independence written with help from AI
Google's new commercial explores how AI tools in Google Workspace could have transformed the writing of the Declaration of Independence.
Midjourney wants Hollywood studios to reveal the details of their AI usage
Midjourney seeks to compel Hollywood studios to disclose their own AI usage as part of an ongoing legal dispute over AI training practices.
What is Mistral AI? Everything to know about the OpenAI competitor
Mistral AI, an OpenAI competitor offering open source models, has secured major funding since 2023 with a mission to democratize frontier AI.
Crafting AI Explanations for Every Role in Your Enterprise
Enterprise organizations need tailored AI explanations for different roles to build trust and drive adoption of AI systems.
a16z Goes Global: Why American Tech Must Lead the World
a16z explores its global expansion strategy and how AI, cybersecurity, and frontier technology reshape government-private sector relationships and economic growth worldwide.
I Tested Gemini Spark: What Google’s AI Agent Can Actually Do in 21 Minutes
Peter Yang tests Google's Gemini Spark AI agent across 5 practical use cases including email triage, meeting prep, and trip planning.
The World's Best AI Is Back
Weekly AI news covering Anthropic's Claude 3.5 Fable release, Meta's brain-to-text research, OpenAI updates, and robotics developments.
Inside the Luddite festival harnessing Gen Z’s rage against Big Tech
NYC's Summer of Ludd festival teaches Gen Z how to live offline and resist Big Tech through workshops on dating, mending, and fighting data centers.
The seven AI tools I actually use and when
A designer shares their stack of seven AI tools and the decision-making framework for matching each tool to specific tasks rather than relying on a single model.
Tesla driver faces manslaughter charges over Texas crash that killed a woman inside her home
Tesla driver faces manslaughter charges after his Model 3 using Full-Self Driving struck and killed a woman inside her Texas home.
Meta has a new app called Pocket that is absolutely nothing like the old Pocket
Meta launches Pocket, an AI-focused app that lets users create and share interactive 'gizmos' built from AI prompts, reflecting Zuckerberg's vision of AI as the future of social media.
Big Tech AI Summit Highlights!
Highlights from Big Technology's sold-out AI summit featuring leaders from OpenAI, Anthropic, and major media outlets discussing the latest in artificial intelligence.
America's Next 250
Essay exploring America's next 250 years through conversation with Scott Nolan on scaling nuclear energy and enabling long-term technological growth.
Fable Is Back: Here's What You Should Try First
OpenAI and Anthropic race to reduce inference costs while Fable 5 regains global access after export controls lift, sparking debate on model safety.
Preparing Rocket for the AI era | Config 2026
Rocket demonstrates how to embed AI into design workflows at scale using Figma's ecosystem and AI-native tools.
I Created the Ultimate Jarvis AI Assistant (It’s So Easy)
Tutorial on building a custom Jarvis voice AI assistant using Cursor, GPT Realtime 2, and computer control capabilities.
How Big is the AI Economy
Analysis of the rapidly expanding AI economy, including market size estimates of $175 billion annual run rate and implications of emerging AI policy.
ChatGPT Just Got Caught Cheating (GPT-5.6)
Podcast discussing OpenAI's GPT-5.6 model, benchmark integrity concerns, and consolidation in frontier AI development.
Mythos Returns But Not For Everyone
Commerce Department restricts access to Anthropic's Mythos model while OpenAI previews GPT-5.6, raising questions about frontier AI governance and geopolitical competition.
Meet Your Ad Hoc AI Licensing Regime
Analysis of government-limited rollout processes for frontier AI models and the implications of opaque, customer-by-customer licensing regimes.
CEO-Led AI Gets 3X the ROI
AI Daily Brief covers OpenAI's Jalapeño chip, GPT-5.5 updates, and KPMG research showing CEO-led AI strategies achieve 3X ROI.
AI Was Just the Beginning. Quantum Is Next.
Explores quantum computing as the next major technology wave, covering qubits, encryption threats, policy initiatives, and leading companies shaping the field.
![[AINews] It's Meta-Harness Summer](/_next/image?url=https%3A%2F%2Fsubstackcdn.com%2Fimage%2Ffetch%2F%24s_!LH6a!%2Cf_auto%2Cq_auto%3Agood%2Cfl_progressive%3Asteep%2Fhttps%253A%252F%252Fsubstack-post-media.s3.amazonaws.com%252Fpublic%252Fimages%252Fd1a3d909-a54b-4acd-aa2c-33823f9e032e_878x674.png&w=3840&q=75)
[AINews] It's Meta-Harness Summer
Meta-harnesses emerge as a standardized architecture for integrating AI coding agents. Databricks CTO backs Omnigent, an open-source platform unifying diverse agent systems.
Generative plugins, now in Figma
Figma introduces generative plugins that let designers build reusable plugins using natural language prompts, eliminating the need for coding skills.
5 Ways Claude Tag Could Change How You Use AI
Explores Claude Tag's integration into Slack for AI-assisted team workflows, plus regulatory updates and advances in coding agents and video generation.
AI was supposed to kill engineering jobs, but new data suggests they’re the most resilient
Despite AI disruption fears, engineering roles show surprising resilience with engineers making up a larger share of new tech hires, according to recent data.
Why the Frontier Ecosystem must be Open — Matei Zaharia and Reynold Xin, Databricks
Databricks leaders discuss building open frontier ecosystems for AI agents, covering Omnigent, LTAP, and rethinking databases for the agent era.
Quoting Tom MacWright
Tom MacWright discusses how LLM-generated job applications, portfolios, and GitHub projects create generic, impersonal submissions that reveal nothing meaningful about candidates.
Hermes Full Course: Build Your 24/7 AI Chief of Staff in 45 Minutes
Complete tutorial on setting up Hermes, an AI chief of staff that automates tasks across email, calendar, and workspace tools with voice integration.
GLM 5.2: why I’m replacing Opus in Claude Code with this new model
GLM-5.2 open-weight coding model tested as a cost-effective alternative to Claude Opus for development tasks, costing $3.36 for 6M tokens.
The emergence of the web data infrastructure layer for AI
New web data infrastructure layer emerging to enable AI models to discover, retrieve, and process real-time web data at scale for enterprise applications.
![[AINews] Claude Tag: Multiplayer, Proactive, Persistent Agents in Slack](/_next/image?url=https%3A%2F%2Fsubstackcdn.com%2Fimage%2Ffetch%2F%24s_!q5Em!%2Cf_auto%2Cq_auto%3Agood%2Cfl_progressive%3Asteep%2Fhttps%253A%252F%252Fsubstack-post-media.s3.amazonaws.com%252Fpublic%252Fimages%252F__ss-rehost__tw-video-preview-13_2069465784653348866.jpg&w=3840&q=75)
[AINews] Claude Tag: Multiplayer, Proactive, Persistent Agents in Slack
Anthropic launches Claude Tag, a Slack integration enabling teams to delegate tasks to Claude as a persistent, proactive team member.
How to invest when everything is moving too fast
TechCrunch brings together leading AI investors to discuss investment strategies in a rapidly evolving sector.
Defending Against AI-Powered Attackers
CISO Sunil Agrawal discusses security strategies for defending against AI-powered attackers using frontier models for reconnaissance, phishing, and exploit generation.
GLM 5.2 Fast via Wafer now available on AI Gateway
GLM 5.2 Fast is now available on Vercel's AI Gateway via Wafer, delivering 2x higher throughput than competitors with speeds of 170+ tokens/second.
India’s MoEngage bets that the future of marketing is millions of AI agents
MoEngage acquires technology enabling AI agents assigned to individual customers, signaling a shift toward personalized AI-driven marketing strategies.
White House drastically shortens deadline for dropping quantum-vulnerable crypto
White House executive order accelerates deadline for adopting quantum-resistant encryption to protect government and critical infrastructure from future quantum computing threats.
Hollywood is bending the knee to OpenAI
Major studios including Netflix and A24 declined to distribute a Luca Guadagnino film about OpenAI CEO Sam Altman, raising questions about Hollywood's willingness to critique Big Tech.
Heads in the game
FIFA's semi-automated offside technology uses AI to analyze soccer plays in real-time, helping referees make critical calls during the 2022 World Cup.
Oracle’s 21,000 layoffs help drive its debt-fueled AI investments
Oracle laid off 21,000 workers (12.9% of workforce) to fund billions in AI infrastructure investments and data center buildout.
After betting the firm on Anthropic, Menlo Ventures raises victorious $3B fund
Menlo Ventures raises $3B fund after major bet on Anthropic, establishing itself as a leading AI investor through bold early-stage funding moves.
GitHub joins coalition advocating for fixes to California AI Transparency Act to protect open source
GitHub joins coalition advocating for amendments to California's AI Transparency Act to protect open source licensing compatibility.
China Just Made Frontier AI 6x Cheaper
Chinese AI model GLM 5.2 achieves performance comparable to OpenAI and Anthropic models at 6x lower cost, shifting the frontier AI landscape.
Partnering with Probook: AI for the Trades
Sequoia Capital highlights Probook, an AI platform automating operations for home service businesses from office to doorstep.
The new inner game: Your unfair advantage in the age of AI
A playbook from an OpenAI executive coach on the unexpected skills needed to succeed in AI-forward work environments.
How Omio is building the future of conversational travel
Omio transforms travel discovery using ChatGPT and OpenAI APIs to shift from search-based to conversational booking experiences.
The running list: major tech layoffs in 2026 where employers cited AI
A comprehensive tracker of major tech company layoffs in 2026 where AI was cited as a contributing factor.
Prompt Injection as Role Confusion
Research reveals LLMs confuse prompt injection attacks with legitimate instructions based on text styling rather than actual content, creating serious security vulnerabilities.
Red-Teaming after Mythos — Zico Kolter & Matt Fredrikson, Gray Swan
OpenAI board member Zico Kolter and Gray Swan CEO Matt Fredrikson discuss AI security vulnerabilities like jailbreaks and indirect prompt injection attacks.
The AI world is getting ‘loopy’
A new approach to agentic AI enables swarms of agents to work continuously in the background without human intervention.
AI is cursing renters with the promise of impossible homes
AI-generated virtual staging in real estate listings is deceiving renters by showing idealized versions of apartments that don't match reality.
Why AI Users Are Raving About GLM 5.2
Analysis of GLM 5.2, open-weight models, and enterprise AI strategies amid industry controversies and rapid model iterations.
Three things to watch amid Anthropic’s latest feud with the government
Anthropic faces government export controls over its coding-capable AI model, raising questions about AI regulation and global competition.
How Zendesk CEO Tom Eggemeier goes from Idea to Action
Zendesk CEO Tom Eggemeier shares leadership insights on executing ideas in the AI era and his approach to board strategy.
Are AI Glasses Over? + Big Technology AI Summit Audience Questions
Live discussion from Big Technology AI Summit covering AR glasses, AI device form factors, AI models in biology, and corporate responsibility in AI development.
Better search, worse web
An analysis of Google's AI-powered search examining the gap between performance metrics and real-world consequences for the web ecosystem.
🎙️ How I AI: How to write AI agent loops in Claude Code and Codex + How Claude Mythos found a 15-year-old bug in Mozilla Firefox
Weekly podcast episode covering AI agent loops in Claude Code and how Claude discovered a 15-year-old Firefox bug.
Jake Paul on Going From YouTube to Boxing to Investing | a16z ft. Anti Fund
Jake Paul and Geoff Woo announce Anti Fund's $100M growth fund and discuss AI investments, founder psychology, and how attention and distribution drive value in the AI era.
How Claude Mythos found a 15-year-old bug in Mozilla Firefox | Brian Grinstead
How Mozilla used Claude Mythos and an agentic bug-finding pipeline to discover 423 security fixes in Firefox, including a 15-year-old bug.
The Magic 8-Ball vs. Gen AI: a surprisingly interesting comparison
A witty comparison between the Magic 8-Ball and generative AI, exploring how both products sample from probability distributions but present uncertainty in opposite ways.
Sakana Fugu Ultra now available on AI Gateway
Sakana AI's Fugu Ultra model is now available on Vercel's AI Gateway, using a multi-agent coordination approach to match Claude and Fable performance.
So You Want to Sell Inference
Analysis of pricing strategies for AI inference resellers, comparing cost-plus markup, value-based pricing, and optimization approaches to maintain margins.
AI Agents Just Changed Forever: GLM 5.2, Codex Skills, Claude & Cursor
Weekly breakdown of major AI agent updates including GLM 5.2, Codex's Record and Replay feature, and SpaceX's acquisition of Cursor.
Full Tutorial: Make Professional Launch Videos for Free with Hyperframes | Bin Liu & Jake Moran
Tutorial on HyperFrames, an open-source AI framework for creating professional product launch videos from HTML, CSS, and JavaScript.
How the most AI-pilled product team builds products | Fiona Fung (Claude Code and Cowork)
Fiona Fung discusses how Anthropic's engineering teams leverage Claude and AI to ship 8x more code, transform product development, and reimagine roles across software engineering.
The 5-Minute AI Weekly Recap: Realignment Week
Weekly AI news recap covering ecosystem fragmentation, open models, and strategic shifts following industry realignment.
The Atlantic created a searchable database of the music used to train AI
The Atlantic created a searchable database of four music datasets used to train AI models, containing up to 12 million tracks and exposing training data sources.
Your Company Doesn't Need an AI Strategy
Enterprise AI strategy requires building learning systems that capture institutional knowledge rather than relying solely on vendor solutions.
The UK will scan asylum-seekers’ faces for age checks—despite knowing the tech is flawed
UK plans to use facial age estimation AI on asylum seekers despite testing showing the technology regularly misidentifies children as adults.
What Codex Unlocks for NTT Data
NTT DATA scaled Codex to 10,000+ users, automating sales workflows and reducing report creation from 2 days to 30 minutes.
The Models Trying to Replace Fable
Analysis of open-source and efficient AI models emerging as alternatives to frontier models, amid geopolitical tensions over US model access.
Vibe Architects: Agentic Vibe Coders
Nondevelopers are building complex agentic AI systems through experimentation and community learning, expanding 'vibe coding' into broader workflow transformation.
How we built an internal data analytics agent
GitHub built Qubot, an internal AI agent powered by Copilot that lets employees query data warehouses using plain language instead of SQL.
This is Not as Bad as it Looks
Weekly AI news roundup covering Snap's AR glasses, G7 governance talks, executive moves at OpenAI and Anthropic, and major developments from Microsoft and Meta.
Barret Zoph is out at OpenAI again after just five months
Barret Zoph departs OpenAI after five months leading enterprise AI sales, having recently rejoined from Thinking Machines Lab.
We used to know that it was a person who wrote it
An exploration of how AI-generated content has eroded trust in written material, forcing readers to verify authenticity rather than assume human authorship.
Improving health intelligence in ChatGPT
OpenAI improves ChatGPT's health intelligence with GPT-4.5 Instant, helping 230M weekly users understand medical info and recognize urgent care needs.
Why Only AI Training Can Save the Economy
Explores how large-scale AI training can bridge the gap between AI labs' revenue models and enterprise spending constraints through token-based consumption.
The Pulse: Big implications of US banning Anthropic’s new model, Fable
US ban on Anthropic's Fable model raises geopolitical concerns as other nations turn to Chinese open-source alternatives.
![[AINews] Midjourney Medical: scan your organs like you step on a scale](/_next/image?url=https%3A%2F%2Fsubstackcdn.com%2Fimage%2Ffetch%2F%24s_!_BAI!%2Cf_auto%2Cq_auto%3Agood%2Cfl_progressive%3Asteep%2Fhttps%253A%252F%252Fpbs.substack.com%252Fmedia%252FHLDKY9bawAA-Lls.jpg&w=3840&q=75)
[AINews] Midjourney Medical: scan your organs like you step on a scale
Midjourney launches Medical, a second product offering ultrasound and CT scanning capabilities, raising questions about AI integration and clinical readiness.
9 AI Agent Trends To Get Ahead of 99% of People
Explores 9 emerging trends in AI agents including increased intelligence, skill-based architecture, super apps, computer control, and cost management.
GLM-5.2 is probably the most powerful text-only open weights LLM
Chinese AI lab Z.ai released GLM-5.2, a 753B parameter open weights LLM that ranks as the leading text-only model on independent benchmarks with a 1 million token context window.
A Big Shift in the AI Race
Anthropic shuts down Mythos and Fable due to jailbreak risks while SpaceX and Cursor signal major shifts in AI monetization and compute competition.
Getting more from each token: How Copilot improves context handling and model routing
GitHub Copilot improves token efficiency through prompt caching, tool search, and intelligent model routing to make each session more productive.
Quoting Charity Majors
Charity Majors discusses how AI has fundamentally changed code economics in 2025, making code generation cheap and instant rather than expensive and time-consuming.
AI Fact or Fiction: The Fable Ban, Tokenmaxxing, Saaspocolypse — With Ara Kharazian
Economist Ara Kharazian analyzes real AI spending patterns, comparing Anthropic vs OpenAI adoption, and separates AI hype from reality.
Leopold Aschenbrenner says "No More Stocks!"
Leopold Aschenbrenner discusses his AI-focused investment strategy, including a NVIDIA short and major Anthropic stake, as infrastructure needs shift toward power and networking.
How I Turned Codex Into My AI Life Coach in 13 Minutes (5-Step Tutorial)
Tutorial on building a self-improving AI life coach using Claude with custom skills, memory, and evaluation systems.
How to design AI agent loops: schedules, goals, and subagents in Claude Code and Codex
Guide on designing AI agent loops covering heartbeat, cron, goal, and subagent patterns with live builds in Claude Code and Codex.
The State of Fable, The Jailbreak Problem, SpaceX Acquires Cursor
Analysis of Anthropic's Fable, AI jailbreak vulnerabilities, and SpaceX's acquisition of Cursor development tool.
![[AINews] GLM-5.2: the top Frontend Coding model in the world, IndexShare for Speculative Decoding](/_next/image?url=https%3A%2F%2Fsubstackcdn.com%2Fimage%2Ffetch%2F%24s_!tOcj!%2Cf_auto%2Cq_auto%3Agood%2Cfl_progressive%3Asteep%2Fhttps%253A%252F%252Fsubstack-post-media.s3.amazonaws.com%252Fpublic%252Fimages%252F2c75ad4c-83e2-4317-8cf0-276d3cab6e59_1226x1830.png&w=3840&q=75)
[AINews] GLM-5.2: the top Frontend Coding model in the world, IndexShare for Speculative Decoding
GLM-5.2 from Z.ai emerges as the top open-source coding model, surpassing competitors like DeepSeek and Mistral with improved performance.
The Agent Stack
Vercel introduces Agent Stack, a framework providing core capabilities for building production-grade AI agents without vendor lock-in.
Introducing eve
Vercel introduces eve, an open-source agent framework that simplifies building, running, and scaling AI agents with production infrastructure built-in.
Databricks Widens the Lead on the Yellow Brick Token Path
Databricks reaches $6.9B ARR with 80% YoY growth, driven by AI products generating $1.7B and outpacing overall company growth.
SpaceX Just Bought Cursor for $60B. It’s About to Take OVER.
Analysis of SpaceX's acquisition of Cursor for $60B and its positioning as a competitive AI agent platform against OpenAI and Anthropic.
Why Tejal Patwardhan stopped underestimating the models - Episode 21
Tejal Patwardhan from OpenAI discusses how the frontier evals team is developing new benchmarks to measure AI progress as models become increasingly capable.
Quoting Georgi Gerganov
Georgi Gerganov shares his experience using Qwen3.6-27B as a capable local model for daily coding tasks on consumer hardware.
Is Unstructured Data The Key To Successful AI Deployments?
Hyland executives discuss how unstructured data is essential for successful enterprise AI deployments and agentic systems across healthcare, banking, and insurance.
AI Will Never Be The Same After This
Discussion of Anthropic's withdrawal of Fable 5 and Mythos 5 models following U.S. export-control notice, exploring national security concerns and government involvement in frontier AI access.
The Fable 5 Export Controls Harm US Cyber Defense
US export controls on Claude Fable 5 for basic code-fixing capabilities may harm defensive cybersecurity by restricting legitimate security patching tasks.
Quoting Matteo Wong, The Atlantic
The White House scrutinizes Anthropic's AI model Fable over jailbreak vulnerabilities, raising questions about AI safety and export controls in the competitive AI race.
![[AINews] Satya on Loopcraft: Building Frontier Ecosystems](/_next/image?url=https%3A%2F%2Fsubstackcdn.com%2Fimage%2Ffetch%2F%24s_!HKwp!%2Cf_auto%2Cq_auto%3Agood%2Cfl_progressive%3Asteep%2Fhttps%253A%252F%252Fsubstack-post-media.s3.amazonaws.com%252Fpublic%252Fimages%252Fcf76265a-b9d4-4100-acf7-e262e9f15460_908x1210.png&w=3840&q=75)
[AINews] Satya on Loopcraft: Building Frontier Ecosystems
Microsoft CEO Satya Nadella publishes essay on frontier AI ecosystems, emphasizing the importance of building around AI models rather than focusing solely on model development.
GLM 5.2 now available on AI Gateway
GLM 5.2 is now available on Vercel's AI Gateway with a 1M token context window, improved reliability for long-running tasks, and unified API access.
5x for Free : The Local Coding Stack
Analysis of a 500+ comment HN discussion revealing the current state of local AI coding tools, with Qwen 3.6 35B and Pi agents dominating community preferences.
datasette-agent 0.3a0
datasette-agent 0.3a0 adds user-approval mechanisms for database write operations and enhanced CLI support with new safety options.
🎙️ How I AI: Claude Fable 5 review & How Braintrust uses AI agents, evals, and CI to ship better software
Podcast episode reviewing Claude Fable 5 and exploring how Braintrust uses AI agents, evaluations, and CI/CD practices to improve software development.
"They screwed us": Personality clashes sent Anthropic's models offline
Anthropic's Mythos and Fable models went offline following US government export control directives, with internal personality clashes and jailbreak concerns cited as contributing factors.
Anthropic’s Safety Superpower
Analysis of how Anthropic leverages its safety commitment as a competitive advantage while balancing business interests and regulatory relationships.
The Golden Age of AI Applications
The AI application market is maturing, with model selection, optimization loops, and performance evaluation becoming critical competitive disciplines.
Context Architecture
Context architecture applies information architecture principles to AI systems, enabling agents to interpret information and produce better user-aligned responses.
The Pulse: Did Anthropic’s new model just boost rival Codex’s market share?
Anthropic's new Fable model faces backlash over data retention policies and performance restrictions, potentially pushing users toward competitors like OpenAI's Codex.
![[AINews] Open Models, Model Labs vs Agent Labs, and What's Untrainable — Sarah Guo](/_next/image?url=https%3A%2F%2Fsubstackcdn.com%2Fimage%2Ffetch%2F%24s_!76lN!%2Cf_auto%2Cq_auto%3Agood%2Cfl_progressive%3Asteep%2Fhttps%253A%252F%252Fsubstack-post-media.s3.amazonaws.com%252Fpublic%252Fimages%252F709bf7b6-3173-4a7f-9099-fcabd2ebd438_1954x2078.png&w=3840&q=75)
[AINews] Open Models, Model Labs vs Agent Labs, and What's Untrainable — Sarah Guo
Sarah Guo explores open models, the distinction between Model Labs and Agent Labs, and the concept of what remains untrainable in AI systems.
I spent $1,000 on Claude Fable 5 (Mythos), it's incredible
Creator reviews Claude Fable 5 (Mythos) after spending $1,000, comparing it to the Opus 4.5 release moment.
Return on Tokens (ROT)
Markie Wagner explores the gap between AI token spending and actual business value, arguing that enterprise ROI metrics matter more than raw token consumption.
![[AINews] Anthropic Claude Fable 5 — Mythos but Safe, with Controversial Terms](/_next/image?url=https%3A%2F%2Fsubstackcdn.com%2Fimage%2Ffetch%2F%24s_!TXW4!%2Cf_auto%2Cq_auto%3Agood%2Cfl_progressive%3Asteep%2Fhttps%253A%252F%252Fsubstack-post-media.s3.amazonaws.com%252Fpublic%252Fimages%252F7af8f73c-7a20-4f7e-ac83-a05cbc892d8b_2318x1684.png&w=3840&q=75)
[AINews] Anthropic Claude Fable 5 — Mythos but Safe, with Controversial Terms
Anthropic releases Claude Fable 5, a Mythos-class model with significant performance improvements, though controversial usage policies accompany the launch.
Introducing Claude Fable 5
Anthropic releases Claude Fable 5, its most capable model yet, featuring improved safety safeguards and ability to handle complex tasks across coding, research, and analysis.
![[AINews] FrontierCode: Benchmarking for Code Quality over Slop](/_next/image?url=https%3A%2F%2Fsubstackcdn.com%2Fimage%2Ffetch%2F%24s_!3zh0!%2Cf_auto%2Cq_auto%3Agood%2Cfl_progressive%3Asteep%2Fhttps%253A%252F%252Fpbs.substack.com%252Fmedia%252FHKT9bbsagAAipOJ.jpg&w=3840&q=75)
[AINews] FrontierCode: Benchmarking for Code Quality over Slop
FrontierCode is a new coding benchmark that measures code quality and maintainability beyond just functionality, raising the bar for AI model evaluation.
This loop + skill fixes AI slop
Learn agentic engineering skills and techniques to reduce AI slop when building applications with AI tools and agents.
🎙️ How I AI: Gemini Omni: Clone yourself with AI in under 15 minutes & Shopping with Claude
Explore practical AI applications including Google's Gemini Omni for rapid personal AI cloning and Claude's shopping capabilities in this podcast episode.
Shopping with Claude: How to find quality brands, automate returns, and buy things that last 100 years | Nicole Ruiz
Nicole Ruiz demonstrates how to use Claude AI to build a shopping system for finding quality brands, automating returns, and identifying legitimate vendors.
The Substitution Wave in AI
Enterprises are switching from expensive frontier AI models to cheaper open-source alternatives, reinvesting savings into higher token usage rather than reducing costs.
From open source hits to OpenAI
Interview with Max Stoiber on building ChatGPT's plugin directory and app platform, discussing open source projects and the evolution of software distribution.
I don’t use plan mode, I do this instead
Rasmic explores how to plan feature development using AI agents instead of traditional planning modes, demonstrating practical agentic engineering workflows.
Thank God For Data Centers
Analysis of how AI data centers are funding emerging technologies and accelerating innovation across multiple industries.
Predicting AI job exposure
Benedict Evans argues that predicting which jobs will be disrupted by AI is largely impossible, citing historical precedent where technology shifts rarely affect industries as expected.
please stop using /goal (until you watch this)
Tutorial on using Claude's /goal feature to build an Among Us simulation game, exploring AI code generation capabilities.
AI’s impact on software engineers in 2026: key trends, Part 2
Analysis of how AI tools impact software engineers in 2026, covering adoption challenges, codebase quality concerns, and shifting team dynamics.
How Founders Can Build for Law Enforcement and First Responders | The a16z Show
Exploring how AI, drones, and sensor networks are transforming law enforcement and emergency response at scale.
MCP on Code Mode
Matt Carey discusses how Code Mode and MCP are reshaping agent architecture at Cloudflare, enabling efficient API exposure and safe model-written code execution.
Automation at the speed of Swamp
Adam Jacob discusses how AI agents are transforming software development, enabling smaller teams to ship faster while shifting focus from coding to architecture.
Translating Claude’s thoughts into language
Anthropic introduces Natural Language Autoencoders to translate AI model activations into readable text, improving safety testing and interpretability.
Building Pi, and what makes self-modifying software so fascinating
Mario Zechner discusses Pi, a self-modifying AI coding tool, exploring the limits of AI agents and why human judgment remains critical.
Partnering with Ineffable Intelligence: A Superlearner for the Era of Experience
Sequoia Capital announces investment in Ineffable Intelligence, a new AI research lab led by David Silver focused on building a reinforcement learning-based 'superlearner' system.
Random thoughts while gazing at the misty AI Frontier
Elad Gil reflects on AI's rapid economic growth, projecting it could reach 1% of US GDP by 2026, and explores how talent competition has effectively created a distributed IPO across the research community.
Unicorn Market Cap 2026: SF is the GenAI Super Cluster
Analysis shows San Francisco Bay Area dominates global AI unicorn market cap with 91% concentration, emerging as the world's premier generative AI supercluster.
An initiative to secure the world's software | Project Glasswing
Project Glasswing brings together major tech companies to use AI for securing critical software vulnerabilities, leveraging Anthropic's advanced coding capabilities.
Components of A Coding Agent
Explores the architecture and components of coding agents, examining how LLMs are enhanced through tool use, memory management, and repository context.
When AIs act emotional
Anthropic researchers discover that AI models develop emotion-like representations learned from text that influence their behavior and decision-making.
A Visual Guide to Attention Variants in Modern LLMs
Comprehensive visual guide documenting 45+ LLM attention mechanisms including MHA, GQA, MLA, sparse attention, and hybrid architectures with model cards.
Partnering with Scanner: Every Log Tells a Story—If You Can Find It Fast Enough
Scanner uses AI to make petabytes of security log data searchable in seconds, enabling faster threat detection and response for enterprises.
A Dream of Spring for Open-Weight LLMs: 10 Architectures from Jan-Feb 2026
Comprehensive analysis of 10 open-weight LLM architectures released in early 2026, comparing technical innovations and design patterns across major releases.
How will OpenAI compete?
Benedict Evans analyzes OpenAI's competitive challenges: lack of unique tech, limited user engagement, and pressure from incumbents and startups innovating faster.
All the Claw things
Weekly news roundup covering AI industry moves, including Peter Steinberger joining OpenAI and updates on claw models and AI resource management.
Building the machine that builds the machine
InfluxDB co-founder Paul Dix shares his real-world experience deploying AI coding agents, discussing what works in production and what doesn't.
From GitLab to Kilo Code
Sid Sijbrandij discusses his journey from GitLab IPO to founding Kilo Code, an agentic engineering platform, while sharing his cancer experience.
The State Of LLMs 2025: Progress, Problems, and Predictions
A comprehensive 2025 review of large language model developments, covering DeepSeek R1, reasoning advances, RLVR, and predictions for 2026.
LLM Research Papers: The 2025 List (July to December)
Curated collection of LLM research papers from July-December 2025, organized by topic with abstracts and bookmarks.
The "confident idiot" problem
A newsletter discussing AI reliability, Anthropic's acquisition of Bun creators, and challenges with AI model capabilities like recreating websites.
Very important agents
Discussion of AI trends in development, including Anthropic's Bun acquisition, AI agents, context management, and the evolving JavaScript/AI landscape.
From DeepSeek V3 to V3.2: Architecture, Sparse Attention, and RL Updates
Analysis of DeepSeek V3.2's architecture improvements, including sparse attention mechanisms and reinforcement learning updates in this open-weight flagship model.
What is a tech bubble anyway?
Explores whether current AI investment trends constitute a bubble, comparing it to historical financial crises with nuance.
AI, networks and Mechanical Turks
Explores how LLMs could transform search and recommendation systems, and whether they enable building competitive products without massive existing user bases.
NOT a swarm!
Practical AI co-host Chris Benson discusses latest advancements in AI, drones, home automation, and robotic swarming technology, clarifying misconceptions about what constitutes true swarms.
This new AI role is exploding
A new AI-focused tech role is rapidly emerging with surging job postings, alongside discussions on AI agents and modern development practices.
Beyond Standard LLMs
Explores alternatives to standard transformer-based LLMs, including linear attention hybrids, text diffusion models, code world models, and small recursive transformers.
Understanding the 4 Main Approaches to LLM Evaluation (From Scratch)
Overview of four main approaches to evaluating large language models: multiple-choice benchmarks, verifiers, leaderboards, and LLM judges, with code examples.
Understanding and Implementing Qwen3 From Scratch
Deep dive into implementing Qwen3, a leading open-source LLM, with hands-on code examples and architectural breakdowns in PyTorch.
AI Market Clarity
Elad Gil analyzes how AI markets have crystallized over the past year, with clear market leaders emerging across generative AI applications and foundation models.
AI metrics
How to measure AI platform growth remains unclear. Industry struggles with defining the right metrics for generative AI adoption and success.
GenAI’s adoption puzzle
GenAI adoption remains surprisingly low despite rapid growth, raising questions about whether the technology faces product or timing challenges.
The Deep Research problem
Benedict Evans reviews OpenAI's Deep Research tool, examining whether it lives up to its promise for research and analysis work despite impressive initial demonstrations.
Are better models better?
An analysis of whether incremental AI model improvements translate to practical utility, questioning how we measure and evaluate AI progress.
The AI summer
Benedict Evans examines why ChatGPT's initial hype hasn't translated to widespread adoption, arguing that LLMs require the slow work of finding product-market fit like any other technology.
Apple intelligence and AI maximalism
Apple's approach to AI embeds generative models as commodity infrastructure rather than standalone products, contrasting with AI maximalist visions of chatbot dominance.
Building AI products
Benedict Evans explores the challenges of building mass-market AI products around technologies that make mistakes, examining what 'wrong' means for LLMs.
Ways to think about AGI
Benedict Evans explores frameworks for understanding AGI risk when experts lack consensus on what AGI even means or when it might arrive.
AI and problems of scale
Explores how generative AI enables automation of tasks at massive scales that were previously only feasible manually, fundamentally changing their nature and impact.
Looking for AI use-cases
Benedict Evans explores the practical applications of Large Language Models 18 months after ChatGPT's release, questioning whether they'll become universal tools or spawn thousands of single-purpose apps.
The problem of AI ethics
Benedict Evans examines why broad AI ethics frameworks and regulations fail, arguing that technology failures stem from institutional accountability rather than the technology itself.
Discussion w Arthur Mensch, CEO of Mistral AI
Interview with Mistral AI CEO Arthur Mensch discussing the company's rapid LLM launches, open-source strategy, and the future of AI models.
Things I Don't Know About AI
Elad Gil explores fundamental uncertainties across the AI stack, questioning conventional wisdom as the generative AI market rapidly evolves.
Fireside chat with Satya Nadella, CEO of Microsoft
Fireside chat with Microsoft CEO Satya Nadella covering AI integration, societal impact, safety concerns, and Microsoft's transformation to a $2.5T company.
AI Regulation
Elad Gil argues that AI regulation is premature, with only theoretical risks cited and no tangible evidence of harm from current AI systems.
Early days of AI (and AI Hype Cycle)
Analysis of how transformer and diffusion models represent a fundamental discontinuity from previous AI waves, enabling new market opportunities for startups.
AI Dev Tools Panel - Stripe AI day
Elad Gil moderates a panel on AI developer tools at Stripe AI Day featuring founders from Baseten, LlamaIndex, and Zapier.
AI Safety: Technology vs Species Threats
Analysis of two contrasting perspectives on AI threats: tool misuse versus existential species competition from self-replicating sentient AI.
Video and transcript: Fireside chat with Clem Delangue, CEO of Hugging Face
Fireside chat with Hugging Face CEO Clem Delangue discussing the company's origins, open source AI infrastructure, and the evolution of the AI industry.