Search papers, labs, and topics across Lattice.
This paper investigates the vulnerabilities of graph foundation models by focusing on their alignment layer, a shared representation that distinguishes them from traditional graph neural networks. The authors demonstrate that a directed perturbation of this representation can lead to significant model collapse at inference time, with OpenGraph being particularly susceptible due to its spectral tokenizer. Their findings reveal that input-space attacks can drastically reduce prediction accuracy, highlighting a critical fragility in these models that is not present in standard graph networks.
A single perturbation can collapse graph foundation models, exposing a unique vulnerability in their alignment layer that traditional graph networks do not possess.
A graph foundation model generalizes across graph domains by mapping every input into one shared representation before any task reasoning. We call this map the alignment layer, the component that separates a graph foundation model from a graph neural network, and we show it is a distinct attack surface that prior work has not studied. We attack it at inference time, with no access to training, on six public models spanning spectral tokenizers, text embedding spaces, and a discrete codebook. A directed representation-space perturbation collapses every model, but at a budget comparable to the representation norm a plain graph network also needs, with one exception: OpenGraph, whose spectral tokenizer collapses at a fifth of that budget, an alignment-specific fragility a plain network does not share and which a same-representation control traces to the tokenizer rather than the decoder. A realizable input-space attack that edits edges, features, or text removes at least half the correct predictions on three of the six models at peak. How much of this fragility an input-access attacker realizes tracks how directly the decoder reads the representation, and not the clean accuracy a task leaves; we measure this carrier gain structurally from the decoder's local Lipschitz sensitivity, and report clean-accuracy headroom as a within-model ordering heuristic that does not survive on realizable attacks.
