Copilot telemetry can now name the user behind agent activity

1. Copilot telemetry can attribute activity to a named user, with five separate capture switches

Worth knowingBuildEurope1 min

Announced on 12 October. Enterprise admins can set telemetry.capture.identity in managed settings to add three OpenTelemetry attributes: user.name (currently the GitHub login), process.user.name (the local OS account) and host.name (the machine name). Separately, five capture classes can be enabled one by one: prompts, responses, toolArguments, toolOutput and policyDetail. Everything is off by default, and managed values override local settings, the SDK and environment variables. The older captureContent setting still works but does not turn on identity. It applies to Copilot CLI, the GitHub Copilot app and VS Code sessions using Agent Host. The post gives no rollout date or plan requirements.

So what: If you run Copilot in a company, decide which classes you actually need for audit before switching any on. Captured data can include source code, commands and file paths, and named employees raise privacy and works-council questions in the EU.

Source: GitHub changelog · Official

2. Copilot for JetBrains lets admins set the default model and stop MCP servers auto-starting

Quick hitBuild½ min

Released on 10 October. Enterprise admins can set any available agent model as the default for new conversations; users can still pick another. A new setting turns off automatic startup of Copilot and Claude MCP servers. Diagnostics get a Fix action that opens inline chat. JetBrains IDE 2025.1 is no longer supported, so you need 2025.2 or later.

So what: If your team is on JetBrains 2025.1, upgrade the IDE before the plugin stops updating for you.

Source: GitHub changelog · Official

3. Ollama 0.40.3 pauses the background model upgrade it introduced in 0.40.2

Quick hitBuildEurope½ min

Follow-up to Monday’s item. The 11 October release pauses the background upgrade of downloaded models and removes warnings from ollama launch claude and ollama launch codex. A 0.40.4 release candidate from 12 October updates llama.cpp and adds installed models to the Codex model picker. Models already upgraded by 0.40.2 still have backup copies on disk, and going below 0.40 means re-pulling them.

So what: If disk space is tight, check for backup copies of models you already ran on 0.40.2, and wait for the final 0.40.4 before updating.

Source: Ollama releases · Official