US Patent Issued to HiddenLayer on Feb. 10 for "Defense of multimodal machine learning models via activation analysis" (Oregon Inventors)

ALEXANDRIA, Va., Feb. 11 -- United States Patent no. 12,549,598, issued on Feb. 10, was assigned to HiddenLayer Inc. (Austin, Texas).

"Defense of multimodal machine learning models via activation analysis" was invented by Amelia Kawasaki (Corvallis, Ore.) and Andrew Davis (Portland, Ore.).

According to the abstract* released by the U.S. Patent & Trademark Office: "An analysis engine receives data characterizing a multimodal prompt for ingestion by a generative artificial intelligence (GenAI) model. The multimodal prompt is processed and fed into a plurality of layers from which an intermediate result of the GenAI model or a proxy of the GenAI model is obtained. The analysis engine, using a prompt injection classifier and the intermediate ...