AI-powered products, tools, and services launched by startups and established companies.
Some satellite companies still have an appetite for boutique launch services
# Boutique Launch Services Remain Critical for Specialized Satellite Operators Dedicated launch services continue to command significant demand in th…
Elevenlabs makes Music v2.5 available via app and API with free and pro tier options
ElevenLabs rolled out Music v2.5, a significant upgrade to its AI music generation platform available now through both the web app and API. The compan…
Final, final, final call for TechCrunch Disrupt 2026 Side Events
# Last Chance to Host an Official Side Event at TechCrunch Disrupt 2026 Organizers have set a hard deadline for proposals to host official side event…
OpenAI's new Agents API gives developers the infrastructure behind Codex and ChatGPT
OpenAI has released its Agents API to public beta, granting developers direct access to the infrastructure that powers both Codex and ChatGPT. The mov…
OpenAI's GPT-Live-1 API lets developers build apps that talk and listen at the same time
OpenAI has released GPT-Live-1, a full-duplex speech API that enables developers to build applications where AI talks and listens simultaneously, mimi…
Muse can shop, write emails, and negotiate prices for users, all through WhatsApp
Meta launched Muse, an AI agent that operates directly through WhatsApp to handle shopping, email composition, price negotiation, and travel bookings.…
Can the US battery market untangle from China?
# Can the US Battery Market Untangle from China? The United States energy storage market is expanding at record speed, driven by massive investments …
Apple Watch’s new AI features are normalizing the idea that technology is always listening
Apple's latest Watch models introduce on-device AI capabilities that transcribe speech and summarize ambient conversations without storing raw audio l…
Suno launches v6 music models built with Warner, BMG, and Believe
Suno, the AI music generation platform, released version 6 of its core models in partnership with three major music labels: Warner Music Group, BMG, a…
AWS is using Qualcomm for AI inference while Qualcomm uses AWS Bedrock to design the chips
# AWS and Qualcomm Enter Symbiotic Chip Partnership for AI Inference AWS and Qualcomm announced a deepening partnership that creates a circular depen…
ChatGPT Images 2.5: Faster, more precise, but not the same for everyone
OpenAI rolled out two new image generation models under the ChatGPT Images 2.5 banner, but the company remains vague about rollout timing and which us…
Batteries just broke another record in the US
Battery storage capacity installations in the United States reached a new record in the second quarter of 2026, with 20.2 gigawatt-hours of new capaci…
Meta debuts its Muse AI agent. Will consumers trust it?
Meta has launched Muse, a personal AI agent designed to handle email, calendar management, payment processing, health services, and other intimate asp…
This founder is teaching chips how to recycle (their energy)
# Chips That Recycle Energy Could Upend Computing's Heat Problem Hannah Earley sees waste heat as a design choice, not an inevitability. The 31-year-…
This founder is making cheaper, cleaner steel
# A Founder Reimagines Steel Production with Lower Costs and Cleaner Emissions Steel production operates on processes barely changed since the 1850s.…
XDOF, just 3 months out of stealth, is in talks for a Series B at a $1.2B valuation
XDOF, a robot data startup, has entered Series B fundraising conversations at a $1.2 billion valuation just three months after emerging from stealth m…
OpenAI developer claims Astra boosted productivity so much it pulled some plans forward by six months
OpenAI developer Thibault Sottiaux says Astra, the company's multimodal AI system, delivered such dramatic productivity gains that OpenAI accelerated …
Google brings AI music generation directly into the Gemini app with its new Lyria 3.5 model
Google has integrated its Lyria 3.5 music generation model directly into Gemini, making AI-powered music creation available to millions of users witho…
Meta's new real-time audio model is the foundation for AI assistants that never stop listening
Meta's Superintelligence Labs released Muse Voice Transcribe, a real-time audio transcription model designed to form the foundation for always-listeni…
Meta is paying to peek at how you use their latest AI model
Meta is running an aggressive pricing experiment with its Muse Spark model. The company offers users a 95% discount on API calls in exchange for shari…
OpenAI launches Astra, its powerful (and controversial) new model
# OpenAI Launches Astra, a New AI Model for Computer and Browser Automation OpenAI unveiled Astra, a new AI model designed to handle computer and bro…
Less than 24 hours to apply for your TechCrunch Disrupt 2026 Side Event
TechCrunch has opened applications for Side Events at its flagship Disrupt 2026 conference, with the deadline set for midnight PT tonight. Companies, …
OpenAI shares prompting tips for GPT-6 Astra including a blocklist of slop words
OpenAI has released a comprehensive prompting guide for GPT-6 Astra, its latest large language model, that goes beyond standard usage instructions by …
OpenAI rolls out GPT-6 Astra to top-tier ChatGPT plans at half the rate of GPT-5.6 Sol
OpenAI has begun rolling out GPT-6 Astra, its latest flagship model, to ChatGPT's paid subscriber tiers with a tiered messaging structure that reflect…
Google’s Gemini Spark can now manage your Google Photos library
Google has extended Gemini's capabilities into photo management, giving Spark users direct control over their Google Photos library. The feature arriv…
Deepseek plans the largest known Huawei chip cluster with 160,000 processors in Inner Mongolia
Deepseek is building the largest known cluster of Huawei chips ever assembled, planning to deploy 160,000 Huawei Ascend-950DT processors in an Inner M…
Nvidia wants your home network to work like a mini data center for local AI
Nvidia is building infrastructure to turn home networks into distributed computing systems optimized for running AI models locally. The company introd…
Pangram's biggest flaw is users turning its scores into public shaming
Pangram, an AI detection tool, faces a credibility crisis as users weaponize its scores to publicly shame content creators accused of using artificial…
The Download: rethinking child safety and fossil-fueled farming
# Child Safety Apps Face Scrutiny as Parents Seek Digital Protection Balance Child-monitoring apps have exploded into a multibillion-dollar market as…
Meta says Muse Spark 1.3 has frontier performance — but its best results come from a model developers can’t broadly use yet
Meta unveiled Muse Spark 1.3, its latest AI coding model, claiming frontier performance at minimal cost. CEO Mark Zuckerberg announced the release as …
Microsoft AI’s MAI-Transcribe-2 undercuts OpenAI, Google and ElevenLabs on price and speed
Microsoft AI released MAI-Transcribe-2 on Thursday, a speech-recognition model priced at 10 cents per hour of audio. That undercuts OpenAI, Google, an…
Meta closes in on the top with Muse Spark 1.3, and undercuts rivals on price
Meta released Muse Spark 1.3, marking its fourth model iteration in five months. The company is pursuing an aggressive strategy of rapid releases pair…
Meta prices Muse Voice Transcribe at $0.18 an hour, with real-time diarization for 20+ speakers: a steal for enterprises?
Meta launches Muse Voice Transcribe, a real-time speech-to-text API priced at $0.18 per hour of audio. Developed by Meta Superintelligence Labs, the m…
Enterprises put non-Nvidia chips 14 points ahead of Nvidia's next-gen GPUs on their evaluation lists
Enterprise customers are actively evaluating alternatives to Nvidia's next-generation GPUs at a rate 14 percentage points higher than interest in Nvid…
US military adds ChatGPT and Grok to AI platform GenAI.mil
The Pentagon has integrated OpenAI's ChatGPT Mil and xAI's Grok for Government into its GenAI.mil platform, expanding the military's access to large l…
Google Gemini's new agent-based video analysis cuts token usage by up to 88 percent
Google has released a new agent-based video analysis system for Gemini that fundamentally changes how the model processes video content. Rather than a…
Anthropic's Claude Fable 5.1 and Mythos 5.1 arrive with a 75% cost reduction for Fable cache reads
Anthropic released Claude Fable 5.1 and Claude Mythos 5.1 on September 1, 2026, marking the company's latest generation of large language models with …
Your files stay put: Perplexity’s hybrid AI keeps confidential data off the cloud
Perplexity rolled out hybrid compute for its Computer platform today, introducing a technical approach that routes sensitive data to local processing …
Runway's Solaris is an AI system that generates software interfaces in real time
Runway has released Solaris, an artificial intelligence system that generates software interfaces frame by frame in real time rather than executing tr…
Instagram admits users often can't tell AI profiles from real people
Instagram is overhauling how it labels artificial intelligence profiles on its platform, effectively admitting that users routinely fail to distinguis…
OpenAI says its ChatGPT ad business hits a $1 billion annual run rate
OpenAI has achieved a $1 billion annual run rate in its advertising business, marking a watershed moment for the company's revenue diversification bey…
China's CXMT makes its first HBM3E chips, closing the AI memory gap
# China's CXMT Achieves HBM3E Production, Narrowing Critical AI Chip Dependency China's largest memory manufacturer, Changxin Memory Technologies (CX…
OpenAI starts charging some customers only when its AI actually works
OpenAI is shifting its pricing model for select enterprise customers toward an outcome-based structure where payment triggers only after the AI succes…
Clipto uses AI to search terabytes of video and is now valued at $250M
Clipto, a San Francisco-based startup, has secured $15 million in Series B funding, pushing its valuation to $250 million. The company develops AI-pow…
Anthropic's Claude Code limit change is a raise on paper but a cut in practice
Anthropic is rolling back a temporary boost to Claude Code's usage limits, resulting in a net reduction despite framing the change as an increase. The…
Musk’s faster path to more gas turbines comes with pollution problem
Elon Musk's SpaceX plans to build an in-house foundry capable of casting turbine blades, claiming the operation will accelerate gas power deployment b…
Nvidia’s AI advantage is moving beyond the GPU
Nvidia's dominance in artificial intelligence hardware extends far beyond raw GPU processing power. The company is now embedding intelligence into the…
Anthropic wants to do for physical hardware what its Model Context Protocol did for software
Anthropic launched the Model Hardware Standard (MHS), a unified interface protocol designed to let AI agents control physical devices with the same ea…
Tokens Aren’t Dollars
# Tokens Aren't Dollars The AI industry has developed a dangerous habit. It measures progress, cost, and complexity through token counts, treating ea…
Cohere Parse 5 loses the benchmark on points. It wins on cost per page.
Cohere released Parse 5 this week, a specialized document parsing model built to handle the scaling problem that plagues enterprise AI workflows. The …
AI benchmarks have a trust problem and Google wants to fix it
Google DeepMind is piloting a double-blind evaluation system designed to eliminate conflicts of interest in AI benchmarking. The test runs on Google's…
Always-on and self-starting AI agents might be OpenAI's next big play
OpenAI is testing "Persistent Mode" for its AI agent Codex, a feature that keeps the system running indefinitely and capable of generating its own fol…
Google's Gemini 3.5 Transcribe turns speech to text in 85 languages while auto-correcting your verbal stumbles
Google released Gemini 3.5 Transcribe, a speech-to-text model that handles 85 languages while automatically removing filler words and correcting verba…
Claude Cowork now runs its own browser inside the desktop app
Anthropic has integrated a built-in browser directly into Claude Cowork, its desktop application for AI-assisted work. The move marks a significant sh…
A startup claims it’s found a drug to make your blood young
Generation Lab has launched an injectable treatment called 1 Generation, positioning itself as a rejuvenation therapy built from two existing drugs. T…
Is Slate Auto’s new electric truck the EV Americans need?
# Slate Auto's Electric Truck Targets America's Reluctant EV Market Electric vehicles remain a niche product in America. EVs account for less than 10…
Google's Gemini Omni 1.1 Flash makes AI video generation cheaper and more flexible
Google has rolled out upgrades to its Gemini Omni 1.1 Flash video model that expand what the system can do while slashing costs and generation time. T…
Alibaba releases Qwen3.8-Flash-Next, targeting "ultimate cost efficiency"
Alibaba's Qwen team unveiled Qwen3.8-Flash-Next, a mixture-of-experts language model that represents a sharp pivot toward extreme efficiency. The mode…
OpenAI's first custom chip "Jalapeño" reportedly beats Nvidia's Blackwell and Rubin in inference benchmarks
OpenAI unveiled "Jalapeño," its first custom-built inference chip, at the Hot Chips conference this week. Early benchmark results show the chip outper…
Perplexity partners with Nvidia to launch Portable Computer, a fully local AI agent with zero token costs
Perplexity is releasing Portable Computer today, an on-device version of its agentic AI platform that eliminates cloud dependency and token costs enti…
Google launches Gemini for legal work to automate contracts and research
Google has launched Gemini Enterprise for Legal, a specialized AI product designed to automate contract analysis, legal research, and document process…
Meta's paid AI agent Hatch launches soon, with a new model called Watermelon due in October
Meta is preparing to enter the paid AI agent market with a product called Hatch, scheduled to launch within weeks. The company will also release a new…
Nvidia says its Groq 3 LPX is four times faster than Cerebras, but the math is more complicated
Nvidia is pushing its Groq 3 LPX inference chip into full production, claiming a four-fold speed advantage over rival Cerebras. The company reports ge…
Zero to Agent in 30 Minutes: Never Type Again with Craig Hewitt
# Zero to Agent in 30 Minutes: Never Type Again with Craig Hewitt Craig Hewitt, founder of Castos, a podcast hosting platform, demonstrated how voice…
Anthropic’s new Claude Tag update lets its Slack agent read the full conversation — and jump in unprompted
Anthropic is shifting its enterprise strategy away from single-user chatbots toward what the company calls "multiplayer AI," positioning AI agents tha…
IBM’s next-gen mainframe chip is the first to run Arm and Z workloads on the same cores
IBM announced a watershed moment for mainframe computing at the Hot Chips conference: a processor that executes both IBM Z and Arm instruction sets on…
Alibaba's Wan3.0 generates AI videos up to 30 seconds long from text, images, and documents
Alibaba launched Wan3.0, a video generation model that converts text prompts, images, PDFs, and PowerPoint presentations into video clips lasting up t…
Cerebras unveils CS-4 with double the performance on the same chip
Cerebras has unveiled its CS-4 AI accelerator, a system the company's CEO Andrew Feldman claims represents the fastest hardware in the industry. The C…
Linkdaze’s smart calendar is built to run a household, not just track a schedule
Linkdaze launches a free smart calendar that moves beyond basic scheduling to manage household operations holistically. The platform integrates AI-dri…
Harvard’s $699 startup bootcamp offers AI avatars of its instructors
Harvard Business School's newly launched HBS Foundry program charges $699 for access to an intensive startup bootcamp that leans heavily on generative…
Inherent, founded by DeepMind alumni, says its AI ‘teammate’ just outperformed Anthropic and OpenAI at replicating research
Inherent, a British AI startup founded by former DeepMind researchers, unveiled Faraday, an AI agent designed to replicate scientific research. The co…
An AI boss fired its first employee but only after humans reminded it of its own rules
Andon Labs deployed an autonomous AI agent named Luna to manage a San Francisco retail store, including hiring and firing decisions. Luna terminated a…
Memory shortage reportedly drives Nvidia AI server prices up about 15 percent
Nvidia is raising prices on its AI server systems by approximately 15 percent due to a critical shortage of DRAM memory components. Samsung, SK Hynix,…
RayNeo's new AI glasses skip the camera, focus on text overlays
RayNeo is releasing a new AI glasses model that deliberately omits two standard features found in most contemporary smart eyewear: a built-in camera a…
Anthropic puts its most powerful model Claude Mythos 5 to work for cyber defense
Anthropic has deployed Claude Mythos 5, its most powerful language model, into cybersecurity workflows through its proprietary Claude Security tool. T…
The Download: threats from space mirrors and credit for AI drugs
# Space Mirror Plans Raise Concerns Over Night Sky Disruption A commercial venture intends to deploy reflective surfaces in orbit to beam sunlight to…
ChatGPT can now send texts for you with new Apple Messages plug-in
OpenAI has released a ChatGPT plug-in for Apple Messages that lets users generate text replies directly within the messaging app. The integration allo…
Runlayer, Rippling drop lawsuits — but the brouhaha is still a cautionary tale for founders
Runlayer and Rippling have ended their legal dispute without settlement payments, marking a truce in what became one of tech's messiest IP clashes. Th…
Europe cancels planned upgrades for Ariane 6 rocket
Europe has shelved planned upgrades to its Ariane 6 rocket, signaling a strategic retreat from ambitious launch vehicle development amid mounting comm…
Reverse-lookup service exposed millions of photos of people’s faces
ClarityCheck, a people-search platform, exposed a database containing over 9 million image files without password protection or encryption. The expose…
Google Discover is getting an AI chatbot-tuned feed
Google is adding natural language customization to its Discover feed, letting users describe content preferences directly to an AI system rather than …
AI Weekly Issue #520: What a week: AI became everybody's decision
Artificial intelligence coverage has fundamentally shifted from tracking technological breakthroughs to monitoring institutional power struggles. No s…
Amazon makes its AI-powered Alexa+ free on Fire TV, no Prime required
Amazon is making Alexa+ available at no cost on Fire TV devices across the U.S., removing the paywall that previously restricted the AI-powered assist…
Amazon aims for delivery drones to reach 500 US neighborhoods by end of 2026
Amazon plans to expand its Prime Air delivery drone service to 500 US neighborhoods by the end of 2026. The company currently operates in a limited nu…
Framework responds to complaints that BIOS update bricks Ryzen 7040 laptops
Framework Computers is replacing some AMD Ryzen 7040 mainboards that were bricked by a BIOS update, even though the affected units fall outside the st…
Meta ran ads for an app promising to nudify female politicians
Meta's advertising platform accepted and ran paid promotions for an application designed to generate non-consensual nude images of women, including US…
The Download: how people really use AI, and Flock’s design choices
AI companies publish usage reports regularly, yet these disclosures reveal limited insight into how people actually interact with their tools. Anthrop…
Wispr raises $280M at $2B valuation as it looks beyond dictation
Wispr, a voice AI startup, has secured $280 million in funding at a $2 billion valuation, signaling investor confidence in its expansion beyond basic …
Satellite operators are in panic mode due to a worsening launch crisis
Satellite operators face an existential threat as launch capacity tightens across the industry. The core problem centers on SpaceX's dominance. Falcon…
Reddit’s AI is turning posts into podcasts and short videos
Reddit is converting text posts into audio and video content through an AI-powered experiment. The platform generates videos that use synthetic voices…
YouTube is changing how it counts views to give the numbers a boost
YouTube is lowering the bar for what counts as a video view. Starting August 24, the platform will count a view the moment playback begins, rather tha…
The Analogue Pocket gets a Supreme makeover in red or gold
Analogue and Supreme have partnered on limited-edition versions of the Analogue Pocket handheld gaming device. The collaboration introduces two metall…
Amazon is trying to crush class-action suits before they get started
Amazon updated its terms of service to require customers to resolve disputes through arbitration rather than class-action lawsuits. The change, announ…
Have a laugh at AI’s expense by roleplaying as a chatbot
A new web tool called "Your AI Slop Bores Me" flips the script on AI chatbots by letting humans role-play as the artificial intelligence instead. The …
Google AI health coach to use Abbott glucose data
Google and Abbott are integrating continuous glucose monitoring data into Google's AI health coaching system through a multiyear partnership. Abbott's…
Anthropic announces watermark detection API that will let third parties detect Claude's AI texts
Anthropic is launching a watermark detection API that enables third parties to identify text generated by Claude, the company's AI assistant. The tech…
OpenAI's Computer History turns your clicks and keystrokes into a searchable ChatGPT memory timeline
OpenAI released Computer History, a Mac application that logs every click, keystroke, and app switch on a user's device, then converts that activity i…
GPT-5.6 Sol goes 14x faster as OpenAI launches Ultrafast mode powered by Cerebras
OpenAI rolled out "Ultrafast" mode for GPT-5.6 Sol, its latest flagship model, delivering output speeds of up to 750 tokens per second. The infrastruc…