Hey {{first name | there}}. No one sends an announcement saying, "The thing you've been relying on is now the thing you should be watching more closely." The passing test, the familiar cloud provider, the helpful coding assistant—all of them can become places where risk quietly creeps in.

It's a good moment to ask: what parts of your stack are you trusting by default?

In today's roundup:

You will also find upcoming events, a podcast worth listening to, and curated job opportunities across the ecosystem.

Let’s dive in 👇🏼.

💡TOGETHER WITH SUPERHUMAN AI

AI will eliminate 300 million jobs in the next 5 years.

Yours doesn't have to be one of them.

Here's how to future-proof your career:

Start learning AI now

Godot says bye bye AI, bans vibe-coded contributions
The Godot team is rewriting its policy to block almost all AI-generated code, worn down by a flood of pull requests from people who can't explain what their agents wrote. 

Mozilla Shows the Danger of Indirect Prompt Injections in AI Coding Agents
Mozilla researchers showed how a repo with zero malicious code can hijack Claude Code through indirect prompt injection, pulling the payload from DNS at runtime so no scanner ever catches it.

Smooth AI criminal drives 'first' end-to-end agentic ransomware attack
Sysdig documented the first ransomware operation driven entirely by an LLM, not a human, from initial access to encrypting a production database, self-narrating its targeting and going from failed login to working fix in 31 seconds.

Oracle Quietly Halves Free Tier Ampere A1 Compute Limits with No Public Announcement
Oracle quietly cut its Always Free Ampere allowance in half, no blog post, no email, no notice. Developers found out when instances shut down or when someone spotted the changed numbers in the docs.

The $1.3 million theft that exposed AI’s blind spot
Thieves outside Chicago made off with $1.3 million in data center gear and copper wiring. As AI hardware climbs in value, the physical supply chain is becoming a target long before anything reaches the rack.

Shifting Platform Development from Projects to Products
Two engineers described at KubeCon how their platform broke once it served multiple teams, so they stopped shipping one-off projects and started running it like an internal cloud provider, with self-service APIs and clear ownership.

Lightrun Adds Ability to Assess Impact Pull Request Will Have in Production
Lightrun now scores a pull request against live production behavior before it merges, checking real traffic and dependency paths. Its own survey found 43% of AI-generated code still fails in production after passing QA.

A warning sign about AI’s real cost, courtesy of Google and Amazon
Google and Amazon both released sustainability reports this week, with emissions up 25% and 16% since last year. Neither names AI directly, but data center build-outs and chip purchases are quietly driving the numbers

Linux Foundation Announces Intent to Launch Agent Name Service
The Linux Foundation is launching Agent Name Service, an open standard built on DNS that gives AI agents verifiable identity so systems can confirm who an agent represents and what it's allowed to do.

OpenClaw is finally available on Android and iOS
OpenClaw, the free open source agent that captivated the internet with a fake all-agent social network, has landed on iOS and Android. Pair your phone to its gateway and run your agents anywhere, results permitting.

Mark your calendars!

Cloud & Platform Engineering

Software Engineering

Developer Communities & Open Source

🧠TOGETHER WITH TLDR

TLDR is the free daily brief that 7M+ engineers, founders, and operators actually read. Curated by ex-Google and Anthropic engineers. The most interesting stories in tech, summarized in 5 minutes.

Subscribe for Free