I may be missing something, but this is exactly what I want as a default. I want the session links there. I see PRs from my coworkers with session links.
I still get to control whether other people can see the session, but I don’t lose it.
I don’t get the “unprofessional” argument. This is attribution. Attribution is professional. If you don’t want it, you turn it off or rewrite the commit messages. If you are having Claude write the commit messages FOR YOU and you are NOT READING THEM then that’s what’s unprofessional. I am honestly kind of shocked and disappointed by how upset people are by this.
Which is actually why I DON'T like the session amended. Ultimately it is an ad placed into the Git log, and noise at that. Regardless of if it exists or not, the professional responsibility for what is contained in that PR/commit falls to the feet of the developer contributing it.
Microsoft previously did this with Copilot, which was correctly met with a very negative response. I don't see why an ad by Anthropic is better than an ad from Microsoft.
Agree, that it is an ad. Also, unless the full session (in a resume-able format, along all session artifact such as intermediate research/docs used to produce the final output) is also included in the commit, the session id is useless.
I presume this is Anthropic trying to their "usage" stats before their IPO.
I'm not sure i see it as an ad. "Sent from my iPhone" is an ad because the iPhone is not a meaningful part of sending the email. Claude is a meaningful part of the commit though if you're using it, so to me it's more like clarity or transparency than an advertisement.
You don't see how a link to "https://claude.ai/.." in the Git log is an ad for Cluade/Anthropic? It feels exactly like "Sent from an iPhone" to me, and has exactly the same value to people reviewing the Git log later (i.e. none).
Keep in mind that if you run across someone else's spam in the Git logs:
- You need a Claude account.
- You need permission to view the session.
The only people who can just follow that link and look at how Claude was used, is Anthropic themselves. Therefore, they can also use it as an additional mechanism to tie sessions back to other IDs/identities. For everyone else this is "Sent from an iPhone" but longer.
But presumably everyone in your company/team is using Jira, so it's not an "ad" because it's a product already used internally. Claude is appending these links to all commits by default, whether or not others on the team use Claude. Those are very different things.
Good counterpoint. We link to GitHub from JIRA, and the reverse. Linking to the Claude session that generated the code seems reasonable...it's a related artifact.
One thing I'm not sure about: what is the utility of linking to the session? What are folks expected to do with that data?
The way I see it, the conversation transcript is the important part. The specific link to claude.ai nor the model itself is important to me. Otherwise we'll all just be linking out to 10 different AI model providers websites to see history. I'd rather it be some open solution.
Models change so much that I don't really see why anybody cares what model was used to generate the code after the fact. It changes every few months and they're non-deterministic so its not like you'll ever "recreate" the same conversation yourself anyways.
I agree. But I think I would also like the OS to put a fingerprint (I.e committed on macOSX in /usr/marc/… while he was watching porn on xxx). This way we can get the full context by default and better understand the PRs.
There has been more than I project where I literally didn’t even look at the code.
I prompted what amounts to verbose “I want X” and then did QA. I’m perfectly happy with an LLM being credited with that code and there being no copyright existing for that code.
For some people that is their life’s work and they want to clutch pearls. Their sole contribution was a half baked idea and they shipped QA out to their “users” but they want to be credited with any rights.
There is definitely a continuum in LLM usage for code, do I agree with an AI blindly stamping every commit, not really.
Do I see LLM generated code as equivalent to me digging a hole with a shovel, it depends. Sometimes there was very little shovelling, other times there was no me involved and you may as well credit the entire hole to the shovel.
Where this will come in is when it comes to copyright and licensing.
If your entire product is LLM generated you sure as hell have no right to place any license on there other than potentially a MIT equivalent license, even that is dubious though but from an organisational standpoint MIT probably will be more widely accepted than saying it is uncopyrightable.
My argument being I would rather know that you didn’t even look at the code, which is what this will achieve, whether I agree with it or not at the very least it will make it easier to sift through the garbage.
Realistically, what do you actually get from viewing people's sessions? I honestly don't really understand why people care.
The specific model used can't be important because theres a new trending one every couple weeks nowadays. The conversation could be wildly out there and have 10+ different iterations of the same thing in it that led to the final product, etc.
It's not like before AI we were asking people to explain every single iteration they went through in their commit body, or to see their entire internal dialogue.
I don't really understand people's hype around needing to see the conversation they had with their AI. As long as the code itself is self-documented and final decisions are relatively well-documented future agents and humans will pick up on it much faster than reading the entire transcripts anyways.
That's not even bringing into the fact that there are people who use multiple models, multiple sessions, multiple harnesses across the same commits/changesets. How do you reconcile those alongside people who could have made human changes in the end?
There’s always been a school of thought that is in this direction. In the pre AI world people discussed whether to squash commits or to not and I was on the squash side because I think that the development artifact should be optimized for reading and just as one does not read a novel in all of its pre-edited forms to read it, one should not do so for code. The cost to grok using that method is too high.
However with LLMs you can have them analyze prior transcripts to identify bug sources and so on. I could see it being used. But all theorycrafting. And perhaps motivated reasoning. I’ve always committed early and often and used bisect to find bugs. That places me right away in the squash commits camp because of the method I use to write code.
People don’t want their incompetence displayed. The session is probably a series of “continue” prompts with basic understanding demonstrated in the initial ask.
I think a lot of people interpret it as trying to foist additional burden on reviewers. As in "here's my PR, I would like to take credit for this idea, it's now on you to review. If you find a problem with it, I expect you to call out exactly what my incorrect assumptions were, which you are able to do because they're all buried somewhere in this Claude session. Let me know what they are and I'll paste them into a new session and submit again."
I think it’s foreseeable that if you use the LLM to commit work you’ve written, it might cause problems with attribution.
In general I assume that attribution to AI is approximate. People copy-paste things out of chats and those don’t get attributed to the AI. And on the flip side, I’ve had AI write a commit, and then I’ve reverted it and written something different by hand (and it got tagged with an LLM session).
The AI attribution is approximate and informative. The commit message and authorship are the more important parts.
You may be concerned that people will think your commits are LLM-generated. I’m sorry that you have to work in that environment, but I don’t work in an environment like that and I don’t share your concerns.
How would it know for certain it was used for a given commit? What if there was no code? What if the approach was different, but some lines were the same?
I think some folks think that using an LLM is a shameful act and this is somehow shaming it and others believe that it ties back to the conversation. These conversations don't have any guarantee of being accessible over the long haul. If there was value in the conversation, the conversation should be somehow captured as well, not just some URLs.
I'm OK with the attribution, but I just didn't like a session URL appearing on a public repo all of a sudden. I was left wondering "did I just leak my private session?"
I don't understand why it isn't opt-in. Or at least a heads-up somewhere.
At this point it's believable to me that Anthropic might not even be aware of what features get added in a given release. It's hard to tell the difference between "they actively are against documenting all of the defaults they keep changing" and "they genuinely don't even pay enough attention to notice when their vibe-coded changes have changed a default". Functionally they're the same, and both would stem from similar (lack of) values, but I think it does kind of matter because it's essentially the difference between explicitly crafting an experience for users versus defining same things they want and letting the vibes end up driving it towards a bunch of user-facing emergent properties that no one has considered.
"Of course they were intentional in adding in an ad to everyone's git log" honestly night be the less cynical take, because the alternative is assuming that they actually look at git logs ever rather than only having Claude deal with it.
I agree that attribution is professional. What I would worry about is people using the link to supply the reasoning behind the change without putting it into the commit message. Then you have the problem of needing to load the session to understand the change, when one would ideally be able to get that context from source control alone.
I use and love openspec and I'm against this session attribution by default setting.
But I do see the value in seeing the conversation that lead to the spec. I make sure that all product decisions go through me prior to the agent writing the spec and the questions I ask and decisions I make could definitely help understand retroactively why something is the way it is.
I guess with the default openspec schema you wouldn't have that but you could customize it or create your own in which the reasoning is encoded fully.
Same. Over relied on and far less clear than alternatives 99% of the time. Claude is a shitty English writer that is only tolerated for it’s coding ability.
If you know “origin” or “source” and those words suffice, using a word that means the same thing but you don’t know is strictly worse for communication.
So if Claude is writing a novel, maybe provenance is the best word to use at times. If its a codebase README or variable name, its usually not.
It is because many "developers" want to pretend that they have read the code and have understood it so they can claim it as their own.
Attribution is indeed professional. If you use an AI to write the code for you and have no idea why you accepted the decisions it made nor can you explain them to another human, that would be clear evidence of acute skill atrophy.
Sure, but it does spook some people like me where I frequently mention context in session like: "it would be better to fix this on the backend, but that is a separate team that moves VERY slowly", etc. And I don't want that accidentally getting broadcast.
Am I misunderstanding this change? This makes it sound like Claude is attaching the entire session content when I thought it was only a link that is exclusively readable by the author, unless deliberately shared?
I can't believe how many people are in favor of it. Like it's so positive in the comments I'd think it's astroturfing. The code is the result of my work however it got done, and if I don't document or explain it well enough then that's a separate issue. Did you want to see my slack messages to my mentor too? Or how about all the stack overflows I read? Maybe I should voice record all the times I swore about Gary the junior dev not documenting something?
It's plenty obvious when the code is AI written it's probably just safe to assume 95% of things have been at least AI assisted. Just look at the drop in SO volume, we're not writing less code we're writing more (rip GitHub uptime).
I disagree with this. The code alone is not the artifact. How you came to the decision to write the code in a certain way is certainly an important artifact of the development process. Otherwise how would your coworkers know whether or not a particular piece of logic is still relevant?
The major problem I see with these are linkrot. Do any of us truly believe these links will work 30 years from now?
Post IPO, many years down the line, a leadership that is eager to cut costs will look at their storage usage and expire 7 year old non-training sessions. Quickest thing to go. I have zero faith in any successful provider getting this right.
Git is supposed to be the durable storage medium, self containing and weathering the passage of time. Now we're littering the repos with a bunch of URLs that are only helpful in the short term.
I don't see any issue with this at all. I noticed it the other day and thought it was convenient. The same way I let the agent tag the PRs/commits with its name. I want to know what was used, in fact I wish it would identify the model used.
I'm not embarrassed by my LLM usage and I want to provide a full audit trail whenever possible to aid myself in the future.
PS: Claude archives old conversations on your computer after 30days. I've started archiving mine to the side as well for potential future use.
I use many tools to develop software and if my IDE would silently start adding "This commit was created with IntelliJ IDEA Ultimate" I'd find it intrusive and obnoxious, too.
If IDEA recorded my manual edits as events and attached a transcript to my commits I don't think I'd mind. If they just tagged it to put their name in that's different.
Knowing which model created the code has more "weight" to it than the editor IMHO which is why I think this is different. Adding the session link is a nice touch.
In my last project I wrote most stuff by hand but also used various Anthropic models, Gemini something and GLM 5.2. By line count the #1 was probably the IDE autocomplete though mostly for boring field initializations and the likes. The "hardest" part was some reverse engineering by GLM that Claude refused but that resultend in zero lines directly. So, how to attribute this?
Allowing tools to insert themselves into the git history only makes sense if you are purely vibecoding with a single model and don't even bother to check and clean it up anymore. Is this truly the default state of software development already? How the times have changed.
I suspect that in the future, lawyers are going to start counseling businesses to implement line-by-line provenance of their source code so they’re prepared for future copyright infringement disputes. Business leaders may also start wanting metrics to understand the impact of AI on development in a more detailed way.
There are hundreds of them for just about every IDE you can think of.
I can't tell if this is a serious question or not, but metaprogramming/generators are decades old concepts that have been common features of basically every IDE since the 90s.
That’s not what we’re talking about, though. That’s mechanical transformation, and isn’t the same thing as code that’s generated on demand from scratch, based on plain-language intent.
Surely you know the difference, right? I assume you’ve used LLMs to generate code by now?
That’s a great question, and is why we have these debates, and have courts to adjudicate disputes. I don’t have all the answers, but the gap between these different behaviors seems pretty wide to me.
Autocomplete doesn’t really work until you start typing. The fact that my phone recognizes that I’m starting to type a word and the IDE offers the most statistically likely choice for me to accept doesn’t make those words its own. Same with autocomplete on a phone or in a word processor when entering sentences.
FWIW the Copyright Office, for its part, recognizes that AI generated code is mechanically generated and its opinion is that it’s not copyrightable.
But that’s not really what we’re talking about here. We’re talking about attribution and when it becomes important. To me the metaprogramming/codegen case is different because of the nature of the mechanism, particularly if you wrote the inputs that the IDE transformed on your behalf.
Nothing at all, I just think its a rather arbitrary distinction, especially when you consider heuristic methods like autocomplete that have been around for ages.
Deletes off your computer. I'm don't know what else they might be doing with it. They keep slimmed down summaries (on your comptuer) I believe but not the full context.
This is probably a controversial take, but I love this. Back in the early days of claude code I had to write a script (well, claude wrote it) to pull out the session id and append it to every commit. It's a lifesaver when you are trying to debug an old commit to pull up the chat session that actually wrote it.
I think its the beginning of a larger trend where Anthropic makes its infeasible to use Claude without Claude getting credit. Not for ethical reasons, but business reasons. Fingerprinting supports this as well.
Im speculating. And these are little things that are reasonable by itself. They are ethical. But I dont think Anthropic wants Claude to ultimately be a tool that produces some output that just gets integrated into some human’s work product. IE they dont care much about authors utilizing Claude for writing their papers or books. They do care about Claude becoming the most prolific published author by orders of magnitude. And not being “plagiarized.”
This annoyed me to the point that I created a python script that could search opencode's database and then a skill that uses the script to find the session I was thinking of. It's nice to be able to say "I think 4 days ago I had an issue with a dropdown in Qt6, find that session." Saves a lot of time and tokens.
Just because it's not in the title and most people won't be scrolling all the way down: its only enabled for web and Remote Control sessions, per a maintainer.
Thing is, if you enable remote control for one session, it becomes globally enabled for all (which I personally prefer as I want to access all my sessions from my phone).
This isn’t quite true. There’s a daemon you can launch with `claude rc` that will let you get at all sessions, but if you just `/rc` in one session only that session becomes available.
Presumably quietly added default-on because everyone had turned attribution off.
Again, if you challenge it “It came from my harness instructions” and obviously telling it not to in memory is about as successful as… any other instruction in memory, about 60% of the time.
I really like working using Claude Code, but Anthropic has a really bad tendency to push out new "features" like this as default-on, through auto-update. That, and their models trying to do everything through bash (including editing files) is enough that I'm seriously looking in to switching harness to pi.
The open models I use everyday also occasionally put the co-signed by claude notices in their commits, because they have been trained on commit logs with these notices present.
We shouldn't polute commit messages with these kind of extra's, they will only lead to more noise. Once all the models append random urls to commit messages, who's to trust these?
If you need attribution I bet that you can find a neater way to tie commit shas to claude uuids.
I like claude code, I use it quite a bit for things, and I don't mind the attribution (I do not plan to deceive people of me using AI or not for a change) and I think a link back to the session is valueable after the fact. At least for me, the session contains a lot of discussion, back and forth, changes from me and more. (Also, if the feature is not on by default, it probably might not exist.)
However, I think claude code needs better change management. It changes quite frequent and I think changes like these should be communicated better in software. (Like a card game may show card changes in game when a patch happened.) It may impact how I do things when creating the commit messages or PRs.
It’s really useful for helping agents navigate history and maintain continuity between sessions. So wouldn’t assume it’s for marketing, it has a lot of utility.
we've started using parallel repos in diff ways for this reason depending on repo purpose, scope etc.
full private repo including session transcripts, etc. agent attribution etc, with public repo that only shows selected subset without sessions ids, some or all agents, etc.
alternatively, we sometimes also do a separate chat only repo or server etc if cross-team agent chat access worth it and don't bother with private / public just disable agents in code repo
have also done some other variants etc. bit of pain to init config, does help with reducing the agent clutter, improve chat security boundaries in case any keys or sensitive env were in any chats if PII / sensitive data checker miss etc.
Kind of crazy how a simple QOL lift (stamping a commit to an agent session, or persisting a session logs on cloud) is considered either innovation or controvertial
Wouldn't think I'd be cheering Antrophic decision, but this is excellent.
They probably were thinking about how that serves them as an ad, not about transparency, but still.
I say they should go one step further and somehow make it harder/inconvenient to turn it off (big tech is good at this already, so...); or, maybe, hide that option behind extra feature under premium subscription or something.
P.S. The comments are pure gold. They don't like they can't fake authorship and have the audacity to be indignant!
As a Claude user this is just a benefit to me. Means I can grab the session and resume if I ever see a commit I want to jump back into, and Claude can read these sessions itself whenever I reference the commit, which helps a bunch.
It might come with advertising benefits for them but expect it’s mostly just because it helps their tools work and that benefits everyone.
Attribution should be opt-in, but they already support it. Via CLAUDE.md where you can specify how exactly AI contribution should be documented. Relying on whatever convention your vendor will establish isn‘t a good idea. Vendors are replaced, project conventions stay, so it‘s better to state your expectations in writing, instead of assuming that someone guesses them right and maintains conformance forever.
Eh, I don't want the tools I use to stamp their names on everything. It's also not uncommon for me to switch between GPT and Claude or different models in the harness if one of them isn't doing well. GPT-5.6 Terra has been doing a significantly better job at writing concise PR bodies and commit messages than Claude's usual multi paragraph word soup.
I don't need "Created by Claude Code 1.2.3, Opus 4.8 inside tmux inside ghostty on macOS ## on Mac Book Pro 2019 by git 123 in /Users/nijave/repo"
If you want session context, commit the session or prompt or add a git note. I think opaque URIs are the wrong approach unless they're to an issue tracker with more details. Proprietary URIs are not easily usable and interoperable with other tools. Commited information should still be valid and useful in 10 years.
I'm also not sure session is the right level of abstraction. I recently setup Claude Code with subscription delegation to opencode harness with z.ai coding plans and opencode go coding plan and it's been working fairly but there is no singular "session"
Hard boundaries there simply don't exist for most workflows, and I suspect if they did, it'd be at the expense of things ~nobody would want to give up.
Interesting. Can you elaborate on why you feel strongly about this? It's easily removable--I don't think your characterization that it "forces" attribution is correct, it just defaults it.
Indeed, the disclaiming of it being human-coded isn't the problem with their commit messages, them using it as a growth hack mechanism is. BTW to say it's attributed to Claude, I don't give that much importance, but I think it's useful to other coders to know what model was used. A simple assisted by AI would mostly handle the plaigiarism problem, because Claude's training data didn't come from Claude.
A lot of people say this is a reasonable feature, and I agree (I would strongly consider having it on), but that's not the defense they think it is.
- Having the feature available seems good.
- Having the feature default on is debatable, but not outside the realm of possibility.
- Having the feature default on, without any kind of release notes that say that it happens and link to the option to disable it is clownshoes "bring back the ghost of Steve Jobs to yell at the product team" type behavior.
On one hand I guess it is fine to automatically tag that the commit was AI-generated. On the other hand, this will happen even if you audit every line and add your own changes too, which means the attribution is misleading.
It’s not meant for saying “this change was done by Claude” it’s meant to connect commits back to the session. Makes it way easier to jump back into a follow up change between sessions as the agent can reload parts of the context.
Anyone thinking this is about tagging AI generated changes and not just making the experience in Claude better is missing the point of the feature.
For a good chunk of my AI coding work, I make my own commits. I add Assisited-by: to them. I glance at the AI commit messages, and though I usually come up with my own messages sometimes I paraphrase. This would just make it a bit more annoying to review the AI commit messages, though I could hide it.
> AI does not enjoy copyright protection. Pretending you created something written by AI is copyfraud.
Which court has made a ruling on this? The only decision I'm aware of was that some Gen-AI artwork couldn't be copyrightable if it was made without substantial input, direction, and scrutiny.
You did not write it, the AI did. Either the work cannot be copyright, or the AI owner (Anthropic) owns the code. Not you. In no case can you own code you didn't write. You guys keep arguing that it enjoys copyright protection without realizing this part. There is no "work-for-hire" default in the law for AI even if it did enjoy copyright. You don't own it. And I am pointing out neither should Anthropic.
That's an opinion from the US copyright office, not a law or a decision made in court. I think it remains to be seen to what extent using generative AI affects the practical application of copyright in larger projects: the office holds that the human-contributed parts still qualify for copyright, and I think there are a variety of possible interpretations and arguments that could effectively allow a largely AI-generated codebase to still be protected by copyright, even if technically big sections of it are not (in a similar manner as e.g. maps are protected, even if the underlying facts are not).
(I would also argue whether it matters in practice the other way: if something was AI generated, it is in principle cheap for someone else to generate it separately anyway: if you're wanting to copy AI generated code then there's probably some part of the ideas in it that are valuable beyond what the AI can output on its own)
(Also, this uncertainty does still mean that it's probably prudent to record which parts of your software are generated and which are not: it's a conceivable outcome that some company winds up in court and manages to argue that the human-generated parts are still covered under copyright, but are unable to point to which parts those are and therefore lose it entirely)
Yes, that's exactly what they were talking about. It says right in that page that sufficient changes can allow them to copyright the work, and it also says that their decision can be challenged.
And once again, you guys make this claim without reading the law.
>several provisions of the Copyright Act imply that it uses the word "author" only to refer to human beings, including provisions (1) vesting copyright ownership "initially in the author"; (2) limiting copyright duration to 70 years after "the author's death"; (3) providing for inheritance of certain rights by the author's "widow or widower" or "surviving children or grandchildren"; (4) requiring a signature to transfer copyright ownership; (5) extending protection to unpublished works regardless of the author's "nationality or domicile"; and (6) defining a "joint work" based on the authors' "intention" to merge their contributions in a certain way.
The law clearly, definitively, is talking about human authorship and nothing else. You are basing this concept that you can copyright AI written code on a law change that has not happened. It is very very unlikely that a law change can/will ever happen, because copyright law in particular is enshrined in international treaties like the 140 year old Berne Convention. This is a major reason why I was very much opposed to the TPP. Every new international copyright treaty is just more cement, making it completely unchangable. If you want to change this law, it will require cooperation from most of the countries in the world to do it.
To be clear, the Copyright Office is providing guidance and a particular (respected) interpretation of the law. Their opinions do influence courts, but they do not constitute law themselves.
Interesting. According to the definition in that Wikipedia link, "copyfraud" is claiming copyright on public domain content.
Does this mean a company can take code from an open source project that used AI, put it into their closed source code, potentially modify it, all without attribution?
Yes, that's exactly what it means. AI works are public domain by default. Copyright law requires the work is created with human creativity. The essential piece is to clearly define what parts are AI generated vs human written. This is an obligation placed on you by the courts, and if you don't do it, it becomes very very hard to argue that you wrote that code.
There is of course, transformative use, and the slew of software packages we used to depend on without acknowledging the original maintainers on every commit.
I don't disagree with you on this. However, it is clear to me that a lot of copyfraud is happening in this industry right now. Just look at all the github repos with a) a licnese.txt and b) a claude.md in the repo.
I still get to control whether other people can see the session, but I don’t lose it.
I don’t get the “unprofessional” argument. This is attribution. Attribution is professional. If you don’t want it, you turn it off or rewrite the commit messages. If you are having Claude write the commit messages FOR YOU and you are NOT READING THEM then that’s what’s unprofessional. I am honestly kind of shocked and disappointed by how upset people are by this.
https://lwn.net/Articles/1091231/
Which is actually why I DON'T like the session amended. Ultimately it is an ad placed into the Git log, and noise at that. Regardless of if it exists or not, the professional responsibility for what is contained in that PR/commit falls to the feet of the developer contributing it.
Microsoft previously did this with Copilot, which was correctly met with a very negative response. I don't see why an ad by Anthropic is better than an ad from Microsoft.
I presume this is Anthropic trying to their "usage" stats before their IPO.
It was an ad, but nobody cared because using iPhone was a point of pride.
Keep in mind that if you run across someone else's spam in the Git logs:
- You need a Claude account.
- You need permission to view the session.
The only people who can just follow that link and look at how Claude was used, is Anthropic themselves. Therefore, they can also use it as an additional mechanism to tie sessions back to other IDs/identities. For everyone else this is "Sent from an iPhone" but longer.
One thing I'm not sure about: what is the utility of linking to the session? What are folks expected to do with that data?
If the tool auto-does it, yes.
Having it in the commit message is handy (not to mentioned I already have it I. The commit message).
Models change so much that I don't really see why anybody cares what model was used to generate the code after the fact. It changes every few months and they're non-deterministic so its not like you'll ever "recreate" the same conversation yourself anyways.
It explains the brevity and any typos. That's how I see it used.
I prompted what amounts to verbose “I want X” and then did QA. I’m perfectly happy with an LLM being credited with that code and there being no copyright existing for that code.
For some people that is their life’s work and they want to clutch pearls. Their sole contribution was a half baked idea and they shipped QA out to their “users” but they want to be credited with any rights.
There is definitely a continuum in LLM usage for code, do I agree with an AI blindly stamping every commit, not really.
Do I see LLM generated code as equivalent to me digging a hole with a shovel, it depends. Sometimes there was very little shovelling, other times there was no me involved and you may as well credit the entire hole to the shovel.
Where this will come in is when it comes to copyright and licensing.
If your entire product is LLM generated you sure as hell have no right to place any license on there other than potentially a MIT equivalent license, even that is dubious though but from an organisational standpoint MIT probably will be more widely accepted than saying it is uncopyrightable.
My argument being I would rather know that you didn’t even look at the code, which is what this will achieve, whether I agree with it or not at the very least it will make it easier to sift through the garbage.
The specific model used can't be important because theres a new trending one every couple weeks nowadays. The conversation could be wildly out there and have 10+ different iterations of the same thing in it that led to the final product, etc.
It's not like before AI we were asking people to explain every single iteration they went through in their commit body, or to see their entire internal dialogue.
I don't really understand people's hype around needing to see the conversation they had with their AI. As long as the code itself is self-documented and final decisions are relatively well-documented future agents and humans will pick up on it much faster than reading the entire transcripts anyways.
That's not even bringing into the fact that there are people who use multiple models, multiple sessions, multiple harnesses across the same commits/changesets. How do you reconcile those alongside people who could have made human changes in the end?
However with LLMs you can have them analyze prior transcripts to identify bug sources and so on. I could see it being used. But all theorycrafting. And perhaps motivated reasoning. I’ve always committed early and often and used bisect to find bugs. That places me right away in the squash commits camp because of the method I use to write code.
I think a lot of people interpret it as trying to foist additional burden on reviewers. As in "here's my PR, I would like to take credit for this idea, it's now on you to review. If you find a problem with it, I expect you to call out exactly what my incorrect assumptions were, which you are able to do because they're all buried somewhere in this Claude session. Let me know what they are and I'll paste them into a new session and submit again."
So no, I'm not going to assume that's what someone meant. I can at least pretend I consider people to be competent.
Also I frequently use an LLM to commit work that I have written. It is just misleading in that case.
In general I assume that attribution to AI is approximate. People copy-paste things out of chats and those don’t get attributed to the AI. And on the flip side, I’ve had AI write a commit, and then I’ve reverted it and written something different by hand (and it got tagged with an LLM session).
The AI attribution is approximate and informative. The commit message and authorship are the more important parts.
You may be concerned that people will think your commits are LLM-generated. I’m sorry that you have to work in that environment, but I don’t work in an environment like that and I don’t share your concerns.
I think some folks think that using an LLM is a shameful act and this is somehow shaming it and others believe that it ties back to the conversation. These conversations don't have any guarantee of being accessible over the long haul. If there was value in the conversation, the conversation should be somehow captured as well, not just some URLs.
I don’t think anyone will find it valuable and I don’t want anyone to read it
I don't understand why it isn't opt-in. Or at least a heads-up somewhere.
"Of course they were intentional in adding in an ad to everyone's git log" honestly night be the less cynical take, because the alternative is assuming that they actually look at git logs ever rather than only having Claude deal with it.
Sometimes I tell Claude ‘don’t trust file X, it was written by a colleague who writes awful tests’. I don’t want that in my commit messages.
It's also, like, kind of the point of using git, isn't it? So putting this in the commit message feels very appropriate to me.
Maybe an ID alone would have been better, not a whole http url.
I just wish they allow for getting inside subscription pricing vs outside subscription pricing easier
https://openspec.dev/
But I do see the value in seeing the conversation that lead to the spec. I make sure that all product decisions go through me prior to the agent writing the spec and the questions I ask and decisions I make could definitely help understand retroactively why something is the way it is.
I guess with the default openspec schema you wouldn't have that but you could customize it or create your own in which the reasoning is encoded fully.
If you know “origin” or “source” and those words suffice, using a word that means the same thing but you don’t know is strictly worse for communication.
So if Claude is writing a novel, maybe provenance is the best word to use at times. If its a codebase README or variable name, its usually not.
Attribution is indeed professional. If you use an AI to write the code for you and have no idea why you accepted the decisions it made nor can you explain them to another human, that would be clear evidence of acute skill atrophy.
This is terrible. I have no desire to read your Claude conversation. I don’t even want to read mine if I do use Claude.
The code is the artifact. There’s no value in trying to make the noisy, verbose conversation one. It’ll likely be dead in future years anyway.
Post IPO, many years down the line, a leadership that is eager to cut costs will look at their storage usage and expire 7 year old non-training sessions. Quickest thing to go. I have zero faith in any successful provider getting this right.
Git is supposed to be the durable storage medium, self containing and weathering the passage of time. Now we're littering the repos with a bunch of URLs that are only helpful in the short term.
I'm not embarrassed by my LLM usage and I want to provide a full audit trail whenever possible to aid myself in the future.
PS: Claude archives old conversations on your computer after 30days. I've started archiving mine to the side as well for potential future use.
I use many tools to develop software and if my IDE would silently start adding "This commit was created with IntelliJ IDEA Ultimate" I'd find it intrusive and obnoxious, too.
Knowing which model created the code has more "weight" to it than the editor IMHO which is why I think this is different. Adding the session link is a nice touch.
Allowing tools to insert themselves into the git history only makes sense if you are purely vibecoding with a single model and don't even bother to check and clean it up anymore. Is this truly the default state of software development already? How the times have changed.
There are hundreds of them for just about every IDE you can think of.
I can't tell if this is a serious question or not, but metaprogramming/generators are decades old concepts that have been common features of basically every IDE since the 90s.
Surely you know the difference, right? I assume you’ve used LLMs to generate code by now?
Is a series of matrix multiplications on a token vector not a mechanical transformation?
Where is the point where you start to anthropomorphize it?
Autocomplete doesn’t really work until you start typing. The fact that my phone recognizes that I’m starting to type a word and the IDE offers the most statistically likely choice for me to accept doesn’t make those words its own. Same with autocomplete on a phone or in a word processor when entering sentences.
FWIW the Copyright Office, for its part, recognizes that AI generated code is mechanically generated and its opinion is that it’s not copyrightable.
But that’s not really what we’re talking about here. We’re talking about attribution and when it becomes important. To me the metaprogramming/codegen case is different because of the nature of the mechanism, particularly if you wrote the inputs that the IDE transformed on your behalf.
The bar for “writes code for you” is very low.
What, specifically, do you have against it?
The sessions are definitely more useful but only if I'm using a single agent from a single provider.
Archives? Or deletes?
I think its the beginning of a larger trend where Anthropic makes its infeasible to use Claude without Claude getting credit. Not for ethical reasons, but business reasons. Fingerprinting supports this as well.
Im speculating. And these are little things that are reasonable by itself. They are ethical. But I dont think Anthropic wants Claude to ultimately be a tool that produces some output that just gets integrated into some human’s work product. IE they dont care much about authors utilizing Claude for writing their papers or books. They do care about Claude becoming the most prolific published author by orders of magnitude. And not being “plagiarized.”
Again, if you challenge it “It came from my harness instructions” and obviously telling it not to in memory is about as successful as… any other instruction in memory, about 60% of the time.
I still have to use Claude Code at work and after getting used to Pi I feel I'm always fighting against it.
Maybe the way I enjoy working with LLMs just works better with Pi.
We shouldn't polute commit messages with these kind of extra's, they will only lead to more noise. Once all the models append random urls to commit messages, who's to trust these?
If you need attribution I bet that you can find a neater way to tie commit shas to claude uuids.
However, I think claude code needs better change management. It changes quite frequent and I think changes like these should be communicated better in software. (Like a card game may show card changes in game when a patch happened.) It may impact how I do things when creating the commit messages or PRs.
> That is a stupid and evil feature.
I can see arguments for "stupid", but "evil"?
full private repo including session transcripts, etc. agent attribution etc, with public repo that only shows selected subset without sessions ids, some or all agents, etc.
alternatively, we sometimes also do a separate chat only repo or server etc if cross-team agent chat access worth it and don't bother with private / public just disable agents in code repo
have also done some other variants etc. bit of pain to init config, does help with reducing the agent clutter, improve chat security boundaries in case any keys or sensitive env were in any chats if PII / sensitive data checker miss etc.
They probably were thinking about how that serves them as an ad, not about transparency, but still.
I say they should go one step further and somehow make it harder/inconvenient to turn it off (big tech is good at this already, so...); or, maybe, hide that option behind extra feature under premium subscription or something.
P.S. The comments are pure gold. They don't like they can't fake authorship and have the audacity to be indignant!
It might come with advertising benefits for them but expect it’s mostly just because it helps their tools work and that benefits everyone.
I don't need "Created by Claude Code 1.2.3, Opus 4.8 inside tmux inside ghostty on macOS ## on Mac Book Pro 2019 by git 123 in /Users/nijave/repo"
If you want session context, commit the session or prompt or add a git note. I think opaque URIs are the wrong approach unless they're to an issue tracker with more details. Proprietary URIs are not easily usable and interoperable with other tools. Commited information should still be valid and useful in 10 years.
I'm also not sure session is the right level of abstraction. I recently setup Claude Code with subscription delegation to opencode harness with z.ai coding plans and opencode go coding plan and it's been working fairly but there is no singular "session"
There absolutely should be a fool proof undefeatable mechanism that identifies or watermarks code that was completely LLM generated.
If you use LLMs, own it.
Hard boundaries there simply don't exist for most workflows, and I suspect if they did, it'd be at the expense of things ~nobody would want to give up.
1: https://x.com/m_herrmann/status/2092579839676948495
- Having the feature available seems good.
- Having the feature default on is debatable, but not outside the realm of possibility.
- Having the feature default on, without any kind of release notes that say that it happens and link to the option to disable it is clownshoes "bring back the ghost of Steve Jobs to yell at the product team" type behavior.
On one hand I guess it is fine to automatically tag that the commit was AI-generated. On the other hand, this will happen even if you audit every line and add your own changes too, which means the attribution is misleading.
Anyone thinking this is about tagging AI generated changes and not just making the experience in Claude better is missing the point of the feature.
https://en.wikipedia.org/wiki/Copyfraud
AI does not enjoy copyright protection. Pretending you created something written by AI is copyfraud.
Which court has made a ruling on this? The only decision I'm aware of was that some Gen-AI artwork couldn't be copyrightable if it was made without substantial input, direction, and scrutiny.
https://www.congress.gov/crs-product/LSB10922
You did not write it, the AI did. Either the work cannot be copyright, or the AI owner (Anthropic) owns the code. Not you. In no case can you own code you didn't write. You guys keep arguing that it enjoys copyright protection without realizing this part. There is no "work-for-hire" default in the law for AI even if it did enjoy copyright. You don't own it. And I am pointing out neither should Anthropic.
(I would also argue whether it matters in practice the other way: if something was AI generated, it is in principle cheap for someone else to generate it separately anyway: if you're wanting to copy AI generated code then there's probably some part of the ideas in it that are valuable beyond what the AI can output on its own)
(Also, this uncertainty does still mean that it's probably prudent to record which parts of your software are generated and which are not: it's a conceivable outcome that some company winds up in court and manages to argue that the human-generated parts are still covered under copyright, but are unable to point to which parts those are and therefore lose it entirely)
>several provisions of the Copyright Act imply that it uses the word "author" only to refer to human beings, including provisions (1) vesting copyright ownership "initially in the author"; (2) limiting copyright duration to 70 years after "the author's death"; (3) providing for inheritance of certain rights by the author's "widow or widower" or "surviving children or grandchildren"; (4) requiring a signature to transfer copyright ownership; (5) extending protection to unpublished works regardless of the author's "nationality or domicile"; and (6) defining a "joint work" based on the authors' "intention" to merge their contributions in a certain way.
The law clearly, definitively, is talking about human authorship and nothing else. You are basing this concept that you can copyright AI written code on a law change that has not happened. It is very very unlikely that a law change can/will ever happen, because copyright law in particular is enshrined in international treaties like the 140 year old Berne Convention. This is a major reason why I was very much opposed to the TPP. Every new international copyright treaty is just more cement, making it completely unchangable. If you want to change this law, it will require cooperation from most of the countries in the world to do it.
Does this mean a company can take code from an open source project that used AI, put it into their closed source code, potentially modify it, all without attribution?