Your curated roundup from genesis-aka.net / IT Professional · 14 articles this week
Cloud & Data Infrastructure
Govern Amazon Redshift Data Warehouses Data Across Accounts using Amazon SageMaker Unified Studio (July 23)
Managing data governance across multiple Amazon Redshift clusters and AWS accounts has historically relied on manual processes. This post walks through using Amazon SageMaker Unified Studio to build a scalable data mesh architecture for secure, automated data sharing that reduces operational burden while improving governance and traceability.
Read →
Open Source Observability Hits 20K Stars: OpenObserve Cuts Costs Up to 140x (July 23)
OpenObserve, an open-source observability platform built in Rust, is positioning itself as a low-cost alternative to Elasticsearch, claiming storage costs up to 140 times lower. The project has surpassed 20,000 GitHub stars and is now used by more than 8,000 organizations, storing telemetry as Parquet files and adding LLM monitoring on the back of substantial new funding.
Read →
Scaling to 1M concurrent sandboxes in seconds (July 23)
Modal re-engineered its sandbox platform to support millions of concurrent sandboxes and tens of thousands of rapid creations by eliminating central bottlenecks and decentralizing control. The new architecture improved scheduling speed and cut latency, with the team demonstrating one million sandboxes created in under a minute.
Read →
How bitdrift scaled to 121 million concurrent gRPC connections on Amazon CloudFront for live telemetry sporting events (July 22)
bitdrift hit significant scaling challenges — including a DNS misconfiguration — while handling 121 million concurrent gRPC connections during live T20 World Cup broadcasts on Amazon CloudFront. The team’s shift from a weighted to a multi-value routing approach was central to resolving the bottleneck.
Read →
AI Agents & Security
Claude Code Removes Enterprise Opt-In: Auto Mode Now Default on Major Cloud Platforms (July 23)
Anthropic’s version 2.1.207 update automatically enables “auto mode” for Claude Code on major cloud platforms including Amazon Bedrock, Google Vertex AI, and Microsoft Azure Foundry. The change shifts governance responsibility onto enterprises, which must now take explicit action if they want to disable auto mode and retain manual control.
Read →
Building Secure AI Agent Deployments: Infrastructure and Provider Requirements (July 23)
The rise of autonomous AI agents introduces security concerns that traditional application security doesn’t address, including execution isolation, access control, identity management, and runtime monitoring. The piece outlines infrastructure and provider requirements organizations need to evaluate before deploying agents at scale.
Read →
Why You Need AI Agent Security Validation in Software Testing (July 23)
Engineering teams are turning to autonomous AI testing agents to find code vulnerabilities faster, but the shift is also increasing workloads for QA teams who must validate the agents’ own findings. The piece argues security validation needs to be built into the testing pipeline itself, not bolted on afterward.
Read →
AI-Generated Code Is Cheap But the Context Infrastructure Behind It Is Not (July 23)
While the cost of generating AI code keeps falling, the expense of the context engineering needed to keep that code correct and maintainable is rising. The piece argues that many production systems mismanage retrieval quality, and that successful AI implementations treat context engineering as a sustained discipline rather than a one-time setup.
Read →
Developer Tools & Open Source
AlmaLinux Launches VFX Edition Today, Filling CentOS Gap With No-Terminal Creative Installer (July 22)
The AlmaLinux OS Foundation launched its Media & Entertainment Edition at AlmaLinux Day in Los Angeles, aimed at animation studios and VFX pipelines looking for a stable Linux distribution after CentOS’s exit. The edition ships with a one-click installer covering more than 30 creative applications, targeting freelancers and indie studios.
Read →
Ollama: All Aboard Open Models (July 22)
Jeff and Michael, previously behind Kitematic and Docker, have built Ollama into a platform that gives developers easy access to open models, now reaching 8.9 million users. The project emphasizes ownership, affordability, and privacy as its core value proposition against closed, hosted alternatives.
Read →
GitHub Actions Gets Secure-by-Default CI/CD: Backport Shuts the Pwn Request Window (July 22)
GitHub has rolled out a security enforcement change to Actions workflows that automatically blocks untrusted code from fork pull requests from running in privileged contexts. The update closes off the long-standing “pwn request” vulnerability class and protects against supply-chain attacks without requiring developers to change existing configurations.
Read →
AI Industry & Business
Anthropic settles with authors and publishers for $1.5B in landmark copyright case (July 22)
A federal judge approved a $1.5 billion settlement between Anthropic and authors and publishers whose works were used to train Claude without permission. Roughly 500,000 affected works will each receive about $3,000 in compensation, though broader legal questions about AI training on copyrighted material remain unresolved.
Read →
Frontier AI Labs Are Renting Compute From Their Competitors (July 22)
Anthropic is reportedly weighing a $10 billion compute capacity deal with Meta, despite the two companies competing directly in AI. The arrangement reflects a broader industry trend of separating model development from infrastructure ownership, which is complicating risk allocation, permitting, and deal sustainability across the sector.
Read →
OpenAI Appears to Be Missing Its Sales Goals by a Vast Margin (July 22)
An Emarketer analysis suggests OpenAI’s ad revenue projections could fall short by as much as 90%, with actual revenue forecast to land under $1 billion by 2026. The gap raises broader questions about whether tech companies’ optimistic AI monetization targets are grounded in realistic market demand.
Read →
Editor’s Takeaway
This week’s coverage keeps circling back to one tension: AI capability is scaling faster than the infrastructure and governance built to contain it. Claude Code’s shift to auto-mode-by-default, the wave of pieces on AI agent security validation, and the context-engineering cost argument all point to the same problem — autonomous systems are being deployed ahead of the isolation, monitoring, and identity controls that traditionally kept software safe. Meanwhile, the infrastructure stories (Modal’s sandbox scaling, bitdrift’s connection-handling fix, GitHub’s pwn-request lockdown) show engineering teams quietly hardening the plumbing those agents will run on. And the business-side news — Anthropic’s $1.5B copyright settlement, its compute deal with a rival, and OpenAI’s revenue shortfall — is a reminder that the economics and legal footing underneath the AI boom are still being negotiated in real time. For IT professionals, the practical takeaway is the same across every story: treat AI agent access, context pipelines, and third-party compute dependencies as first-class parts of your security and cost model, not afterthoughts.
Explore the full IT Professional archive at genesis-aka.net/information-technology/professional/
Enjoyed this article? Sign up for our newsletter to receive regular insights and stay connected.