Infrastructure Exits and the AI Authenticity Tax
This week's stories share one theme: trust under transition. Two infrastructure providers are leaving markets, GitHub is rebuilding its storage layer for agent-scale load, and two pieces ask how much of what you read was written by a person. Quad9 pulls its DNS resolver out of France Quad9, the pri

This week's stories share one theme: trust under transition. Two infrastructure providers are leaving markets, GitHub is rebuilding its storage layer for agent-scale load, and two pieces ask how much of what you read was written by a person.
Quad9, the privacy-focused public DNS resolver, announced it will withdraw from France. The decision ends a long-running legal dispute over whether the resolver must block sites at the DNS level.
If you point systems or users at Quad9, plan a migration. The resolver IP 9.9.9.9 may appear in router configs, deployment scripts, or internal docs, and each reference needs review.
Network admins and anyone serving French traffic should pick a fallback and compare filtering policies. Privacy guarantees differ between providers.
Source: Quad9 (DNS Resolver) to Exit France
Indian certificate authority eMudhra announced it will exit the global SSL/TLS certificate business. The press release sets out the wind-down for existing certificates.
TLS certificates are plumbing until one expires on a production hostname. If your fleet includes eMudhra-issued certificates, inventory them now and schedule reissues with another CA ahead of the deadline.
Ops and platform teams should audit their issuance pipelines. Any automation hardcoded to eMudhra needs a second provider before the first renewal fails.
Source: eMudhra to exit global SSL/TLS certificate business
GitHub's Bug Bounty team published a profile of researcher @r31n. It covers their research approach, favorite bug classes, and advice for people entering the field.
The value is in the methodology, not the specific bugs. Reading how one researcher triages scope and decides where to dig sharpens your own code review habits.
If you write code that runs on the internet, this is worth twenty minutes. If you are considering bounty work, the advice section maps what triage actually looks like.
Source: How bug bounty sharpens one researcher's security skills
GitHub engineering wrote up the migration of 99.99999% of repositories to Azure. The remaining fraction waits on per-reference quorums, and object storage now lets reads scale independently of the primary copy.
This is the substrate your pull requests sit on. Moving git objects to object storage is how GitHub plans to absorb the read load that coding agents generate.
Platform and infrastructure engineers should read the quorum design closely. Everyone else can note it and move on, since the user-visible surface should not change.
Source: The bridge to agent-scale Git infrastructure
Martin Fowler's fragment responds to Gergely Orosz's "no AI" essay competition, where entrants used AI anyway. Fowler's point: when readers suspect AI, they stop investing effort, and assume the whole text is machine-written.
That assumption is the tax. A draft that is 90% yours can be discounted to zero by a reader who spots the seam.
Fowler names real caveats, including assistive uses like a colleague's Parkinson's second brain and translation for non-native writers. The honest summary: if you ship text for humans, a machine-perceivable AI voice can cost you the reader before your argument lands.
GitHub talked with three Java Champions about how they use Copilot day to day. The pattern across all three: the tool accelerates the work, and the human stays responsible for the result.
That framing is the useful part. These are experienced developers with deep Java context, so their acceptance criteria for suggestions are high.
Java teams evaluating Copilot should read this before any benchmark. The practical detail is how they verify output, not how fast they type.
Source: How 3 Java Champions are using GitHub Copilot for increased productivity
GitHub documented how to continue a Copilot app session in VS Code. The session carries over with its conversation, code changes, and working context intact.
Agent sessions die most often at the handoff between surfaces. If the app-to-editor bridge works as described, you can start a question in the app and finish it inside your repository.
Anyone using the Copilot app should try one handoff this week. If the context survives, it changes where you choose to start a session.
Source: How to continue a GitHub Copilot app session in VS Code
These write-ups are researched and published with no paywall, sponsor, or tracking. If one saved you an afternoon, a small tip keeps them coming.