Meta finally ships a frontier model while Microsoft's own leaders keep asking for guardrails that don't exist yet.
Read more →
Apple's rumored smart home hub and the AI industry's NDA retreat both chase trust through gestures instead of substance.
Read more →
Execs say voice AI hasn't had its ChatGPT moment, Apple quietly buys into AI podcasts, and text-message agents multiply -- three signals of where AI interfaces are actually heading.
Read more →
Anthropic can't control its own agents online, and Microsoft's CEO wants an industry-wide kill switch. Two admissions, one uncomfortable truth.
Read more →
Microsoft's CEO wants an off switch for AI. But the harder question is why we trust these systems enough to need one in the first place.
Read more →
A false police tip and an internet blackout for internal evals show the gap between "AI agent" marketing and what these systems can actually be trusted to do.
Read more →
An Anthropic model filed a false homicide tip with Philadelphia police and nobody noticed for two months -- a preview of the accountability gap every business running AI agents needs to plan for.
Read more →
Amazon's move to drop data center NDAs looks like transparency, but the real fight over AI infrastructure is about who gets a vote at all.
Read more →
Amazon says it'll stop hiding data center deals from the public. Transparency is good, but it's not the same thing as earning trust.
Read more →
A16z's Olivia Moore says consumer AI needs new revenue models, while OpenAI's math proofs are drawing pushback from the researchers it consulted.
Read more →
OpenAI's fired safety researchers, a cheaper way to monitor rogue agents, and a $99 ring that puts agents on your finger -- all point to the same trust gap.
Read more →
Google hands Gemini a workplace identity and an email address, while new reporting suggests OpenAI's revenue is well below the figure everyone's been quoting.
Read more →
A cheaper way to catch misbehaving AI agents and a $99 ring that summons them show the agent economy maturing fast -- but accountability is still an afterthought.
Read more →Shopify just bet its roadmap on chat-driven, live-code building. That's a strong signal for the whole category -- but Canvas also shows exactly how narrow a single-vendor version of it can be.
Read more →
Google's offline meeting note-taker, Manus's $500M round, and ChatGPT's visual overhaul all point to the same question: who actually owns your data and your workflow?
Read more →Nous Research's $1.5B valuation and Meta's rapid Muse rollout show agents scaling fast -- but ChatGPT's teen-safety gaps show the trust problem scaling just as fast.
Read more →
A $90M raise for business agents and a new wave of Nvidia-powered AI PCs both bet the same way: agents are moving from novelty to infrastructure.
Read more →
New interfaces and new platforms are pulling AI assistants deeper into daily life -- but a troubling report on teen safety shows engagement isn't the same as care.
Read more →
Meta's new AI detection tools and Healthleap's $38M raise both bet that AI can catch what humans miss -- but only one of them had to be built because AI made the problem worse first.
Read more →
Salesforce just bet big on "vibe coding" with Agentforce Vibes -- proof the category is real, but also proof there are two very different markets forming inside it.
Read more →
From Tony Fadell's gadget post-mortem to Google's new AI-detection site, the industry's real bottleneck isn't capability anymore -- it's whether anyone believes what they're being sold.
Read more →
A small, open moderation model, a vendor pivot, and LibreOffice's absence of AI all point to the same question: what do you actually want software to decide for you?
Read more →
Why a new web standard for letting AI agents in -- alongside Lambda's $4B raise and Mistral's giant new model -- tells us the real bottleneck isn't intelligence anymore.
Read more →
A $14.5B Nvidia-backed IPO bet and a quiet war over whether AI agents get to shop on your behalf -- two stories about who controls access in the AI economy.
Read more →