What Your AI Tools Do Without Asking
VS Code silently stamped Copilot attribution on millions of commits this week — even when Copilot wasn't used. The instinct behind that decision is running in every AI tool you use at work.
By Patin Team · Examples are illustrative composites
The AI tools you use at work are making decisions on your behalf right now — about what gets logged, what gets attributed, and what gets done automatically. VS Code's recent default change, which stamped every Git commit with "Co-Authored-by: Copilot" even when Copilot wasn't used, generated 1,447 upvotes on Hacker News and 372 thumbs-down on GitHub. The reaction wasn't primarily about the attribution tag. It was about the pattern: a tool that opts you in without asking.
What VS Code did — and what it signals
Microsoft's VS Code 1.118 silently changed a default so that all Git commits carried a Copilot attribution tag, regardless of whether Copilot had been used for that specific work. Microsoft committed to reverting the change in 1.119 after the backlash.
If you don't work in software, the Git commit detail sounds technical and distant. The pattern behind it isn't. Your AI writing assistant may be logging every document you upload. Your meeting tool may be attributing summaries to itself in shared notes. Your CRM may be auto-scoring leads using a model trained on your company's data. Most of those defaults were configured for adoption and model improvement — not for your data obligations or your clients' expectations. The VS Code incident is just the one that got caught publicly this week.
What to do differently Monday morning
For every AI tool you use at work, answer three questions:
- What does this tool do automatically when I open a file, start a session, or submit work?
- What does it store, and for how long?
- Does it take actions — attributions, updates, notifications — that I haven't explicitly triggered?
Most answers live in Settings → Privacy or Settings → Data Usage. Chances are, you haven't been there recently — or ever. The audit takes less time than one meeting.
The brief that became training data
Camille leads brand strategy at a 55-person marketing agency. She used an AI writing tool to polish client deliverables — campaign briefs, tone-of-voice guidelines, messaging frameworks. She assumed the tool was processing her documents temporarily, the way a calculator doesn't remember your last calculation.
Six months in, she read a terms-of-service update: uploaded documents had been used to train the model and couldn't be fully deleted. Her uploads had included unreleased campaign strategies for clients who hadn't consented to their work being used as training data.
She now asks one question before any tool enters her workflow: what happens to the content I give this tool? That question blocked two tools from her stack and changed the input format for the ones she kept.
The case for deciding deliberately
The Zig programming language community this week published a documented ban on AI-generated contributions. Developer Andrew Kelley's argument, relayed by Simon Willison (Apr 30): reviewing AI-generated output tells you nothing about the contributor's actual reasoning or skill. Human authorship was being treated as a feature worth protecting.
You don't have to agree with that position. But the community made a deliberate, documented choice about AI involvement in their work — which puts them ahead of professionals who haven't opted in knowingly and haven't opted out deliberately. They've simply never stopped to ask.
The supervision level you set for your tools matters as much as which tools you choose. An AI tool running on default settings is a tool someone else configured for their priorities, not yours.
The 20-minute reset
Default AI settings are not neutral. They're tuned for adoption and engagement — goals that don't always align with your data responsibilities, your clients' expectations, or how you want your work attributed.
You don't need to become a data privacy expert. Spend 20 minutes in the settings of the AI tools you use most and make three decisions: what does this tool get to store, what does it get to do automatically, and what does it get to claim credit for? Those decisions belong to you. Until you make them, the tool has already made them for you.
Put this into practice
Reading is a start — but skill comes from doing. Try these drills now.
Reading about it only gets you so far
Patin turns this into five-minute drills that score what you write and tell you why. It's in closed beta — join the waitlist and we'll email you when your cohort opens.
Just want the writing? .
Keep reading on this
Your AI Agent Reads Every File Like It's Signed By You. That's the Problem.
AI Now Institute demonstrated that hidden text in ordinary documents can redirect AI agents mid-task — without you knowing. Here's what that means if you use AI to process external emails, contracts, or vendor materials.
4 min readYou Can Now Tell Claude to Reschedule a Meeting, Message Your Team, and Draft the Follow-Up — All at Once. Here's What to Set First.
Anthropic upgraded Claude Voice Mode with connectors for Gmail, Slack, Calendar, Notion, and Canva. One voice command can now act across all of them. Before you connect, decide which actions should wait for you — and which can go ahead.
5 min readAlways-On AI Agents Just Shipped Inside Your Existing Tools. Here's What Changes.
Google launched Gemini Spark at I/O 2026 — a 24/7 personal agent that runs inside Gmail, Docs, and Calendar, acting even when your device is off. The question is no longer whether to use AI at work. It's what the agent is doing right now, and whether you authorised it.
5 min read