In partnership with
Welcome, AI enthusiasts
For years the guessing game around OpenAI's first gadget pointed to one thing, a phone that would finally take on the iPhone. What the company is building instead is stranger and quieter, a screenless speaker meant to live in your home, shaped by the team that once built the iPhone. Let's dive in!
In today’s insights:
OpenAI's First Device Won't Be a Phone
Google Deepmind CEO Hassabis Wants to Brake AI Before AGI Arrives
Anthropic Drops Claude Pro FREE for US Teachers
Read time: 5 minutes
LATEST DEVELOPMENTS
AI HARDWARE
🤖 OpenAI's First Device Won't Be a Phone
Source: Sfstandard
Evolving AI: OpenAI's first device will be a screenless, portable speaker built as your live-in AI companion.
Key Points:
Instead of the long-rumored phone, OpenAI is opening with a screenless speaker it calls a new kind of home computer.
A camera and facial recognition let it recognize who is nearby, and it acts on its own, even nudging its owner toward an earlier bedtime before a big morning.
The speaker is expected to cost $200 to $300 and ship in early 2027, even as Apple sues to block OpenAI's hardware.
Details:
Apple's former legendary designed Jony Ive, responsible for designing the iphone is behind the design of the device as OpenAI bought his startup io Products for $6.5 billion and hired more than 400 former Apple employees, including designers who built the iPhone. A rechargeable battery lets the device move room to room while a camera and sensors read its surroundings, and its voice runs on GPT-Live, a new mode that can listen and talk at once. It leans on personal details such as emails to build a picture of its owner's life.
Why It Matters:
ChatGPT reaches most people only when they open an app, so a model that lives in the home and speaks like a person never switches off. For anyone who brings one in, the draw is a helper that learns the household well enough to ease the work of running it. That closeness is the catch, since it works only by letting an always-on system into the private corners of home.
TOGETHER WITH DATADOG
🔐 LLM Observability Best Practices
Evolving AI: 4 Key Insights for Scaling LLM Applications.
LLM workflows can be complex, opaque, and difficult to secure. Get the latest ebook from Datadog for practical strategies to monitor, troubleshoot, and protect your LLM applications in production. You’ll get key insights into how to overcome the challenges of deploying LLMs securely and at scale, from debugging multi-step workflows to detecting prompt injection attacks.
FRONTIER AI BRAKE
🛑 Google Deepmind CEO Hassabis Wants to Brake AI Before AGI Arrives
Source: Google I/O
Evolving AI: Google DeepMind CEO Demis Hassabis is racing to build AGI and now wants a way to slow that race down.
Key Points:
Hassabis says AGI is probably only a few short years away, a moment he calls pivotal for humanity.
He wants the US to build a Frontier AI Standards Body, modeled on FINRA, that would vet the most advanced models before they are released.
The body could go further and coordinate a slowdown across the frontier labs if the risks grew too high.
Details:
Wall Street has long policed itself through FINRA, and that is the model Hassabis reached for. The body would run mostly on industry funding, with a board of independent technical experts and open-source representatives. It would probe frontier models for dangerous cyber, biological and deception skills. Labs would share their models up to 30 days before release, voluntary at first, then required to pass before a model could deploy in the US market. The rules would cover every frontier-class model, whatever country it comes from and whether it is open or closed.
Why It Matters:
Google's own models would have to clear this body before they reached you, and so would every rival. A model could sit in testing while evaluators examine what it can really do, and the upgrades you have gotten used to would arrive on someone else's clock. The choice of when your next AI lands would no longer sit only with the frontier labs.
10x the context. Half the time.
Speak your prompts into ChatGPT or Claude and get detailed, paste-ready input that actually gives you useful output. Wispr Flow captures what you'd cut when typing. Free on Mac, Windows, and iPhone.
Evolving AI: Verified US K-12 teachers can now claim Claude Pro wrapped in a teaching toolkit.
Key Points:
Teachers get a full free year by verifying their US K-12 status, with sign-ups open through June 30, 2027.
Learning Commons ties lesson plans to academic standards across all 50 states and evidence-based curricula.
Claude never trains on educator conversations, and student data sits under FERPA-aligned terms.
Details:
Anthropic is opening the program to individual US teachers first, with a dedicated school and district version still on the way. Cowork can run a scheduled job like the weekly parent meeting without a teacher at the keyboard, while each account pairs Claude Pro with teaching skills for lesson planning and differentiation, plus connectors to curricula like Illustrative Math and OpenSciEd.
Why It Matters:
Claude turns those skills and curricula into a standards-aligned lesson or quiz in minutes, then hands the review and the final call back to the teacher. The unpaid prep that swallows a teacher's evenings falls away, and for millions of students a frontier model quietly enters the daily rhythm of the classroom.
👀 Click on the image you think is real
QUICK HITS
💼 Monzo co-founder Tom Blomfield left Y Combinator to join Anthropic's compute team.
⚖️ Hachette, Elsevier, Cengage, and author Scott Turow sued Google, alleging Gemini was trained on pirated books and journal articles.
🔗 Claude Artifacts now support public sharing and multiplayer editing, plus creation via Claude Tag in Slack.
🏛️ NY Gov. Hochul signed the nation's first statewide moratorium on new hyperscale data centers, pausing permits for up to a year.
📈 Trending AI Tools
🚀 Sanebox - AI-Powered Inbox Assistant which automatically filters out unimportant emails*.
🧠 AirJelly - Local Mac agent that watches your apps and captures tasks automatically.
⚙️ Kuberns - AI agent that deploys and scales your apps with zero DevOps.
🦖 Osaurus - Open-source AI agents that run 100% locally on your Mac.
*partner link