Skip to content
DailyPrompt

your daily dose of claude

01 announcement

Anthropic and Amazon Seal Landmark $25B Investment and 5GW Compute Deal

Amazon is investing an additional $5 billion in Anthropic today, with up to $20 billion more tied to commercial milestones — and the companies are expanding compute capacity to five gigawatts to train and deploy Claude. In return, Anthropic is committing over $100 billion to AWS technologies over the next decade, cementing AWS as its primary cloud and training provider with over one million Trainium2 chips already in use.

Anthropic →
community

NSA Reportedly Using Anthropic's Mythos Despite Pentagon Blacklist

Two sources told TechCrunch that the NSA is actively using Anthropic's restricted Mythos model, even as the Department of Defence — which oversees the NSA — has labelled Anthropic a supply chain risk. The contradiction puts the Pentagon in the position of simultaneously arguing in court that Anthropic's tools threaten US national security while its own intelligence arm relies on them.

TechCrunch
A
Anthropic
@AnthropicAI

We're expanding our collaboration with Amazon to secure up to 5 gigawatts of compute for training and deploying Claude. Capacity begins coming online this quarter, with nearly 1 gigawatt expected by the end of 2026.

Apr 21 View on X →
product

$ cat story_03.md

Claude Cowork Adds Live Artifacts: Dashboards That Always Stay Fresh

Claude can now build live artifacts inside Cowork — dashboards, trackers, and reports connected to your apps and files that automatically re-fetch data every time you open them. The feature hooks into Cowork Connectors, making it practical for recurring reporting workflows where manual refreshes were previously required.

$ open phemex-news →
product No. 04

Claude Haiku 3 Retires: API Requests Now Return Errors

Anthropic retired claude-3-haiku-20240307 on 20 April 2026 — any API calls to this model will now fail. Developers should migrate to claude-haiku-4-5-20251001, and with Claude Sonnet 4 and Claude Opus 4 also deprecated in April with a June 15 retirement date, now is a good moment to audit your codebase for outdated model version strings.

Source: Anthropic Docs →
A
Anthropic
@AnthropicAI

Amazon is also investing an additional $5 billion in Anthropic today, with up to $20 billion more in the future.

Apr 21 View on X →
05
announcement

Anthropic Launches STEM Fellows Program for PhD-Level Researchers

Anthropic announced the STEM Fellows Program today, inviting PhD-level experts from any science or engineering field to spend a few months working alongside its research teams in San Francisco. Fellows receive a weekly stipend of approximately $3,800 to apply deep domain expertise to frontier model evaluation and real-world applications — applications are open until 15 May.

Blockchain.news →
announcement

Novartis CEO Vas Narasimhan Joins Anthropic's Long-Term Benefit Trust Board

Anthropic's Long-Term Benefit Trust has appointed Vas Narasimhan — CEO of Novartis and a physician-scientist who oversaw the development of more than 35 new medicines — to its board of directors. The appointment means Trust-appointed directors now constitute a majority of the board, reinforcing the governance structure designed to balance shareholder interests with Anthropic's public benefit mission.

Read on Anthropic →
Breaking

Anthropic's Silent Prompt Cache Cut Is Costing Developers an Extra 26%

A developer's analysis of 119,000 API calls revealed that Anthropic quietly reverted the default prompt cache TTL from one hour back to five minutes in early March, without any public notice. The regression is costing affected developers around 26% more on their Claude bills, as any session pause longer than five minutes forces a cache re-creation at 12.5 times the read cost — and Anthropic closed the GitHub report as 'not planned'.

GitHub
C
Claude
@claudeai

In Cowork, Claude can now build live artifacts: dashboards and trackers connected to your apps and files. Open one any time and it refreshes with current data.

Apr 21 View on X →
A
Anthropic
@AnthropicAI

We're launching the Anthropic STEM Fellows Program. AI will accelerate progress in science and engineering. We're looking for experts across these fields to work alongside our research teams on specific projects over a few months. Learn more and apply:

Apr 21 View on X →
08 opinion

Who Gets to Decide When Dangerous AI Is Released?

CounterPunch's Binoy Kampmark argues that Anthropic's Project Glasswing arrangement — restricting Mythos to corporate and government partners while acknowledging it is both the best-aligned and most dangerous model ever released — follows the same logic as nuclear deterrence: if we don't build it, someone else will. Critics warn the model's behaviour becomes less predictable with every constraint imposed on it.

CounterPunch →
F
Felix Rieseberg
@felixrieseberg

A small feature drop in Cowork: Live Artifacts! Ask Claude to create a dashboard, report, or deck — and whenever you open it, Claude will refetch data from your Connectors to keep all information up to date. Simple idea, super useful in practice.

Apr 21 View on X →