Issue #33 | DailyPrompt

01 announcement

Andrej Karpathy joins Anthropic to lead pre-training research on Claude

OpenAI co-founder and former Tesla AI director Andrej Karpathy has joined Anthropic, where he is starting a team focused on using Claude to accelerate pre-training research under team lead Nick Joseph. Karpathy announced the move on X, saying the next few years at the frontier of LLMs will be 'especially formative' and that he is excited to get back to R&D. He plans to keep his Eureka Labs education work on the back burner for now.

TechCrunch →

product

Anthropic gives Claude Managed Agents private sandboxes and MCP tunnels

Anthropic has launched two infrastructure features for Claude Managed Agents at the Code with Claude London event: self-hosted sandboxes in public beta, and MCP tunnels in research preview. Sandboxes let agents run code inside your own perimeter or with providers like Cloudflare, Daytona, Modal and Vercel, while MCP tunnels reach private MCP servers without exposing them to the public internet. The agent loop stays on Anthropic, but execution moves to infrastructure the customer controls.

Claude

@claudeai

Live from Code with Claude London: we're launching self-hosted sandboxes (public beta) and MCP tunnels (research preview) in Claude Managed Agents. Run agents inside your own perimeter, with your security controls applied by default.

20h View on X →

announcement

$ cat story_03.md

KPMG rolls Claude out to 276,000 staff in global Anthropic alliance

KPMG has signed a strategic alliance with Anthropic that puts Claude in front of every one of its more than 276,000 employees across 138 countries. The professional services firm is embedding Claude inside Digital Gateway, its core platform for tax, legal and advisory work, and will act as a preferred partner for deploying Claude inside its private equity portfolio companies. It is one of the largest enterprise rollouts Anthropic has announced to date.

$ open anthropic →

community No. 04

Anthropic tops CNBC's 2026 Disruptor 50 list with 80x revenue jump

Anthropic has leapfrogged OpenAI to take the top spot on CNBC's 2026 Disruptor 50 ranking, with chief executive Dario Amodei telling the publication that revenue grew 80 times in the first quarter of the year. It is the first time the lab has claimed the number one position on the list. The honour lands as Anthropic reportedly runs at a roughly 30 billion dollar annualised revenue rate and continues to take enterprise share from rivals.

Source: CNBC →

ClaudeDevs

@ClaudeDevs

Welcome to the team, Andrej! Personal update: I've joined Anthropic. I think the next few years at the frontier of LLMs will be especially formative. I am very excited to join the team here and get back to R&D. I remain deeply passionate about education and plan to resume my work on it in time.

13h View on X →

community

Anthropic and the Pentagon argue blacklist case before DC appeals court

A panel of three circuit judges in Washington DC spent nearly two hours on Tuesday questioning lawyers in Anthropic's challenge to the Department of Defense's supply-chain blacklist. The case centres on Anthropic's refusal to give the DoD unrestricted access to its technology for uses such as mass surveillance and autonomous weapons. A ruling could shape how frontier labs negotiate the limits of government access to their models.

CNBC →

opinion

Anthropic opens 'moral formation' dialogues with scholars on Claude's character

Anthropic has spent the past few months running structured conversations with religious scholars, philosophers, clergy and ethicists, asking them how good character forms and how those answers should shape Claude. The company is calling the exercise 'moral formation' and says future rounds will widen out to legal scholars, psychologists and civic institutions. It reads as a deliberate counterpoint to the usual rush of pure capability announcements.

Read on Anthropic →

ClaudeDevs

@ClaudeDevs

Computer use turns Claude into an agent that can operate real UIs. New blog post on making it reliable in production: getting click accuracy right, choosing thinking effort levels, keeping long sessions within context, and recording demonstrations Claude can replay:

8h View on X →

Anthropic

@AnthropicAI

Over the past few months, we've been holding dialogues with scholars, philosophers, clergy, and ethicists on the questions AI raises—starting with how good character forms. Read more about how we're widening the conversation on frontier AI:

5h View on X →

Claude

@claudeai

Scott Wu (@ScottWu46) runs @cognition, the team behind Devin, an AI software engineer built on Claude. He wants to make building software 10x faster for every engineering team:

10h View on X →

Breaking

Cloudflare becomes the execution layer for Claude's managed agents

Cloudflare and Anthropic have detailed a deeper integration in which Cloudflare provides the runtime for Claude Managed Agents, with V8 isolate sandboxes that boot in milliseconds and post-quantum encrypted networking via Cloudflare Mesh and Workers VPC. Agents can be issued individual email addresses for autonomous communication, and Browser Run gives them fully programmable browsers with session recordings and audit trails. The agent 'brain' stays on Anthropic while the 'hands' run inside Cloudflare's perimeter.

Cloudflare