I am no longer letting Claude Code add itself as Co-author in my commits

For the past twelvemonth aliases possibly more, I’ve been definitive astir adding Claude Code as co-author for each perpetrate (and besides wanted others to do the same) that I utilized it for. And my reasoning for it was simple: arsenic a reviewer aliases anyone other reading the code, they should cognize that I utilized an LLM to build it.

However, now it’s evident and moreover expected that utilizing LLMs is the measurement forward and you have to usage it astatine activity unless you want to spell retired the door. Side projects is wholly upto you and we person had aggregate instances of folks publicly rejecting its usage off-late.

A carpenter whitethorn motion their work. But they don’t adhd the sanction of the saw they used to trim wood, do they? Adding the LLM arsenic co-author feels similar.

And while I’m still very overmuch utilizing LLMs for some activity and side-projects I am not going to adhd them arsenic co-authors to my commits. My reasoning is rather simple: if the activity is nether my name, I presume afloat work for it. A fewer years ago I’d ne'er person added Google aliases StackOverflow arsenic co-authors. But I would have added a nexus to station that helped and I’ll proceed to do truthful now.

In my view, adding an LLM arsenic co-author gives you a measurement for you to shirk responsibility for mistakes – moreover if a small bit. You tin say, oops sorry this happened because it did its point and it was not careful. And that is exactly the behaviour I want to defender against. If a portion of activity is nether my name, I own it. Fully.

To beryllium clear, location person ever been bugs. Both earlier and aft LLMs. So our attitude towards them shouldn’t alteration conscionable because the instrumentality nether the hood has changed. We still request to stay responsible for verifying the activity earlier we ship it.

However, arsenic pinch everything there’s scope for nuance present arsenic well. It whitethorn be okay and astir apt amended to usage LLMs arsenic co-authors aliases moreover arsenic the primary authors for a fewer of the scenarios that travel to mind:

• You’re evaluating different models for akin tasks and want to easily distinguish betwixt them for comparison
• An LLM is creating the PR and you are the reviewer

What astir attributions and disclaimers for AI generated work? If you’re concerned being transparent to anyone seeing your work, either create an AI specific relationship that commits pinch your sanction aliases adhd a large disclaimer to your project’s readme that this is AI slop and location beryllium dragons. The aforesaid measurement you’d have done for thing of a hack that you haven’t afloat tested yourself. But for thing serious, please return afloat ownership of it.

Appendix

I woke up this greeting and saw this post connected HackerNews and it reminded of my ain connected Bluesky from earlier this week. And I felt powerfully that I needed to get this portion retired correct away.

Early connected I believed we should beryllium adding LLMs arsenic co-authors to commits we make. My stance connected this has now flipped. I deliberation LLMs arsenic co-authors makes america shirk work for our changes. You wouldn’t person added Google aliases Stackoverflow arsenic co-author previously. So you shouldn’t do it now.