Track papers across labs, topics, and trends.
Search papers, labs, and topics across Lattice.
Department of Computer Science , The University of Texas at Austin
3
0
1
4
Adversarial data poisoning can hijack protein design models, forcing them to generate specific, unintended protein sequences with remarkable reliability, even when the model's core weights remain untouched.
Model-generated DNA can now be traced with perfect accuracy, even after single-base edits, using a watermark detectable without knowing its precise location or strand.
The biomedical significance is that, under the stated threat model, a served genomic model can process an encoded sequence without exposing plaintext queryderived activations to the compute provider.