Docker, Inc

Running AI agents is easy. Running them safely is the hard part. Docker is covering that challenge end to end at the Data Science Dojo's Agentic AI Conference (virtual) - from governance to hands-on implementation: - Panel: Governing Autonomy: Policy, Control, and Accountability in Agentic AI Systems - With Tushar Jain 📅 April 6 | 9:15–10:00 AM PT - Tutorial: Securing Coding Agents: Sandboxes, Guardrails, and Real-World Attacks: Michael Irwin 📅 April 7 | 9:10–10:00 AM PT - Workshop: Building Secure Sandboxes for Autonomous AI Agents: Oleg Šelajev 📅 April 9 | 9:00–11:00 AM PT If you’re moving from experiments to real agent systems, this is the full picture. Join from anywhere (virtual!), register → https://bit.ly/41m6Jt7

Axios was compromised for three hours this week, and that was enough. A hijacked maintainer account deployed backdoored versions carrying RATs attributed to Lazarus Group, the same playbook behind TeamPCP, Shai-Hulud, and GlassWorm. The pattern is consistent: stolen creds poison trusted packages, poisoned packages steal more creds, and the loop is accelerating. Docker's CISO on what actually failed and what engineering teams should change now. Read → https://bit.ly/41Nqpq2

If you’re using AI agents today, you’ve probably hit this already. It’s not the model that’s slowing you down - it’s everything around it. Running multiple agents, keeping track of what they’re doing, and moving beyond your laptop without losing control. In this conversation, Warp CEO Zach Lloyd and Docker President and COO Mark Cavage break down why the real bottleneck is infrastructure - and what needs to change to make agent workflows actually scale. Watch the full video → https://lnkd.in/dkn7HFEB 

🚀 Gemma 4 is now on Docker Hub! Docker Hub hosts a curated collection of cutting-edge AI models, from lightweight models to high-performance LLMs, all packaged as OCI artifacts and ready to run. Now part of that lineup, Gemma 4 is designed to support a wide range of applications, and Docker Hub makes it simple to pull and run. Learn more: https://bit.ly/4v7cdpr

Docker Model Runner now supports the NVIDIA DGX Station GB300. The hardware is powerful but what makes it practical is the developer experience: pull, run, and iterate on larger models using the same Docker workflow you already know and trust. Learn more here:

Agents in YOLO mode. Safely. Sandboxes are now standalone, giving any agent or claw workload a fully isolated environment to run in. Inside a microVM, agents can install packages, run Docker, and execute freely - without touching your host. No shared kernel. No cleanup required. Full autonomy, with real isolation. Just `brew install docker/tap/sbx` and go. Get started: https://lnkd.in/guhFJY_C 

If you’re one of the growing number of developers trying to balance agent autonomy with control - catch Docker's Tushar Jain at the Future of Data & AI: Agentic AI Conference panel. This discussion will cover how to design governance for agentic systems - from architectural control mechanisms and human-in-the-loop oversight to auditability, compliance readiness, and scaling autonomy responsibly across real-world environments. Governing Autonomy: Policy, Control, and Accountability in Agentic AI Systems 📅 April 6, 2026 - Virtual! ⏰ 9:15–10:00 AM PT Register below ⬇️

New episode is here in the Global AI Community 's Made for Dev Docker series. Oleg Šelajev breaks down how to secure AI-driven development workflows in practice: • Docker Hardened Images to reduce CVE noise • VM-based Sandboxes to isolate agents • Secure API key handling via network proxy • MCP guardrails for controlling tool access Useful for experienced devs looking to level up, or anyone getting started with Docker in agent workflows. Watch → https://lnkd.in/gGDPqCcJ

New episode of Ship Happens is out. Per Ploug Krogslund sits down with Pipedrive's Director of Engineering Infrastructure, Kristjan Elias, to talk about what infrastructure teams are navigating right now: balancing cost, reliability, and security, deciding when AI belongs in production, and where local models make sense vs external providers. They also get into the bigger shift underneath all of this - how platform teams build trust, manage change, and support the business without over-engineering for perfection. Catch the full conversation → https://lnkd.in/gcsY8Bdm