The microVMs probably don't have "two co-operating workloads are able to communicate" in their threat model. The main concern would be if a malicious payload was able to affect or extract data from any other VM.

Ask the least smart LLM you have access to (I tried gemini lite flash) to make a list of ways for processes in microVMs to use a sidechannel attack to talk to each other and you'll get quite a list (including obviously cache hits/misses). Heck a smart CS junior would come up with it. It's well studied!

Both workloads being malicious and trying to establish a hidden data channel between them probably makes this attack a lot easier than conventional spectre/meltdown where it's trying to target ordinary processes that are not intentionally trying to be loud.

Hidden data channel between cooperating processes is not quite as alarming, but it may have some implications for malware command and control.

properly implementing things like iommu, amd-vi, svm/hvms and such things more granularly can easily avoid this kind of crap. it costs a lot of processing and performance which is why people skip over it or dont do it.. or maybe they dont read CPU manuals anymore. idk..

its relatively easy to reset caches etc. between contexts. so them it becomes the game of what a context is.

Makes sense when you think about it- CPU cache would be a layer / vulnerability level up from the microVM layer

Also a somewhat common exploit pattern for low level stuff (along with dram and cpu cache is not all that different)

Seems like a win to me. Their microvms can now be hardened in the lab, not in an emergency :)

The ask matters for the "rogue model" story, but I agree it should not be the whole story for isolation. Two cooperating tenants who both want a channel is a much easier problem than one VM stealing from another that is not playing along, and several people downthread already draw that line. What still bothers me is how little hardware you needed once both sides were willing: a shared cache was enough, and that design is the default, not an exotic bug. If the threat model is only "uncooperative neighbor," this demo does not break it. If the product claim is "these VMs cannot communicate," it does.

I’m not sure I see a real threat here.

If you’re running an untrusted workload, you’re already running it.

Whether it can communicate with another untrusted workload, even though it should be separated, meh.

Unexpected? Probably. but I’m not sure it’s a security risk.

The distinction is interesting, but for a different concern. One thing is to have rogue models breaking stuff, other is to have them following orders, and breaking stuff with success.

This means cloud models can have some guardrails and follow them most of the time (we could model their authorizations like a benign employee with permission to break glass during incidents). But self hosted models can avoid that and help malicious people do advanced attacks, which is a shame.

In principle I'm all for self hosted models, to avoid risks of AI centralization, but they also have other types of risks.