Author of the blog post here: just for the record, I did try replaying against other models. I replayed from GPT 5.5 to 5.5-mini. It "worked" (replay not rejected) but mini didn't hand me the plaintext of the reasoning. I'm not a jailbreaking expert, so I didn't really "push" on it very hard. I also didn't pursue Opus/Haiku for reasons I don't really recall. These folks took it all the way. I'm bummed it didn't work for me, but having these folks prove the channel out is a very good consolation prize.

PS Here's a conversation I had with GPT 5.6 about the paper differences. https://chatgpt.com/share/6a7b64b4-ec0c-83ea-a9d2-ab1f1a1dfe...

If you didn’t allow it, you wouldn’t be able to change models in the same conversation, as key parts of the context would be lost.

Wouldn’t surprise me if the providers just remove that ability and lock the model once the conversation starts.

Fair (I haven't been using the encrypted-reasoning systems, though this is common in open ones - I'm kinda surprised it's an option in encrypted ones too), though what they're doing here is cross-user replays in addition to cross-model.

AFAIK no provider guarantees compatibility of reasoning traces, even in the same model generation, and we've in practice seen most of the big LLM APIs throw errors indicating incompatibility (at least transiently) when switching models. The only stable solution right now is to just throw away reasoning traces whenever a model is switched.

100% guaranteed that this research just forced this to happen now.

Sucks.

It’s already patched according to the authors. Details were not specified.

or add some metadata and don't allow downgrading.

There seems to be an obvious choice to make here, should you give the users to decrypt and use the COT that they did not generate themselves?

This is only required if you want users to be able to share things with everyone and you are going for the simplest implementation.

If not you could try to keep a record of keys associated with a user, then when a new request comes in look through to see if the user has a valid key to decrypt the COT.

For explicit shares, just add the key used in that one conversation to the users valid keys. For global shares use the global keys. But that's adding more complexity to the system.

It’s about being able to change models mid-task. For example, I want to be able to plan using Fable but implement the plan using Sonnet, and that won’t work if this is implemented.

Or even my fable credits run out mid task and need to switch back to opus >.<

Prior to LLMs I never considered that I might have to make a resource-usage decision between hiring Star Trek's Data vs. his stupider brother B4...

https://memory-alpha.fandom.com/wiki/B-4

Star Trek is a post-scarcity society, those problems don't exist there unless you're in the middle of a crisis and on emergency power.

LLMs briefly seemed like this too, after subscriptions made the SOTA models too cheap to meter, but before they walked back on that and introduced quotas...

Actually, that brings up a good reason they can't fix it. Fable falls back to Opus when the topic is too "unsafe". That behavior requires traces than can move between models!

For plan it's relatively easy, just make the plan the artifact. The point is to ingest knowledge with one model and use it in another, and that is not necessarily easily expressible in natural language.

I suspect that there are companies with internal proxies that load-balance across keys, and they didn't want to break that when adding encrypted reasoning.

I really don't understand why server-side storage of the trace isn't a viable approach here, with only a unique key flowing to the client and back. Does it have something to do with how backend load-balancing works?

Makes no difference. There is a policy as to whether to allow use of a reasoning trace in a given context. Whether that trace originates from authenticated ciphertext or a backend database is basically irrelevant.

Good point, thanks.

Yes, this storage would be growing exponentially making the disk space and latency problems harder (add the disaster recovery/backups). I think the choice of using client side is not too bad if you ensure that its secured properly. Also the company can excuse itself from the liability of storing sensitive data on its servers, thats a big deal in itself to be compliant for enterprise audits

1. The down side is that it cannot be used across the clients even for the same user

2. Using the same encryption key was a bad choice here, a per user key would have solved this issue for sure.

Having thought about this a little more, it's clear that server-side storage is not compatible with Zero Data Retention (ZDR). However, in non-ZDR settings, it seems likely that the providers are capturing all that data anyway?

> a per user key would have solved this issue for sure

It would have helped with PII leakage, but not with plain-text trace extraction attacks, right?

Per user encryption key ties it with the user session (assuming you do authentication properly), no one else can access it. User being able to see the information is not really an attack vector in this case.

The compliance rules at times are outdated and people skirt around them by following the worded rule instead of the intent.

Encrypted state cookies solve real problems (server-side storage, latency, scaling) and are not the problem. The problem is insufficient binding of some of a session's encrypted state cookies and others -- insufficient binding of some session state to other session state. Here we have HTTP encrypted state cookies for identifying authenticate user IDs and maybe for identifying sessions / chats, while the reasoning traces are also encrypted state cookies but not HTTP cookies, and the latter are somehow not sufficiently bound to the former.

The fix is to either have per-user or per-session keys for encrypting reasoning traces, or write the user ID / account ID and maybe also session ID into the plaintext of the reasoning trace _then check that that matches the ones in the HTTP cookies when decrypting the traces_.

That’s incompatible with zero data retention and so you’ll lose a lot of enterprise customers.

I believe its a result of chasing the speed and happy path. The intent was to make the data available for reuse on the backend side for the same user session, it kind of makes sense. Making it available across the models is most likely by design.

The flaw is that the data is not strictly tied to user session, making the session data hijacking a lot easier.

1. Its a security issue.

2. Publicly available sessions make it much worse

I'd lean yes; switching between models, even mid conversation, seems to be encouraged by harness vendors. No reasoning makes this much harder.

You can switch models mid conversation so it makes sense that it is allowed. Otherwise you would lose all the reasoning traces.