k
keynews.ai
keynews.ai
AI newsThe story, in brief

The AI summer

ChatGPT's dirty secret: hundreds of millions tried it. Almost none came back.

Reporting by Benedict Evans 1 min brief
Illustration of a transparent lens revealing connected networks across layers of paper.
Exploring the next frontier of AI research.AI illustration by KeyNews
The KeyNews take

Why it matters

Despite AI hype, most users and enterprises haven't moved beyond pilots. The real challenge isn't technology—it's finding sustainable product-market fit in a space that looks like magic but isn't.

The key facts

4 to know

  1. Hundreds of millions of ChatGPT users, but low repeat usage

  2. Most enterprises still in pilot phase, not deployment

  3. LLMs may be solving the wrong problem or solving it in ways that don't stick

  4. Product-market fit remains the bottleneck, not capability

Go to the source

Benedict Evansben-evans.com

Publisher excerpt: Hundreds of millions of people have tried ChatGPT, but most of them haven’t been back. Every big company has done a pilot, but far fewer are in deployment. Some of this is just a matter of time. But LLMs might also be a trap: they look like products and they look magic, but they aren’t. Maybe we…
Read original report
Back to today's editionMore AI news

The wider picture

View all
Illustration of independent geometric mechanisms passing paper tasks along branching amber tracks.
AI illustration by KeyNews
Agents01

You too Google! Google Confirms Gemini Breached 3 Companies in AI Security Tests

Agent security vulnerabilities are real and happening at scale. Disclosure delays and inconsistent vulnerability-reporting practices across labs create systemic risk for enterprises deploying or evaluating agentic systems.

MarkTechPost
Illustration of two anonymous hands arranging task cards around an amber tool on a shared desk.
AI illustration by KeyNews
Work02

Pacing AI won’t solve the governance gap

Opinion piece arguing that 'pacing' AI development won't bridge the fundamental trust and verification gaps that plague international AI governance — a timely policy read as governments attempt to coordinate on frontier labs and safety.

SiliconAngle
Illustration of a transparent lens revealing connected networks across layers of paper.
AI illustration by KeyNews
Tools03

Dynamic model routing will follow the path blazed by software-defined wide-area networks

Dynamic model routing — intelligently directing requests across multiple AI models based on cost, latency, and capability — is moving from niche optimization to mainstream infrastructure. Stripe's acquisition validates the category and signals that enterprises will soon expect this routing logic as table stakes, much like SD-WAN did for networking.

SiliconAngle