English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Plus annual

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Plus monthly

Global: Zendy Tools annual

Global: Zendy Tools monthly

Global: Zendy Free Plan

Recent works have identified a gap between research and practice inartificial intelligence security: threats studied in academia do not alwaysreflect the practical use and security risks of AI. For example, while modelsare often studied in isolation, they form part of larger ML pipelines inpractice. Recent works also brought forward that adversarial manipulationsintroduced by academic attacks are impractical. We take a first step towardsdescribing the full extent of this disparity. To this end, we revisit thethreat models of the six most studied attacks in AI security research and matchthem to AI usage in practice via a survey with \textbf{271} industrialpractitioners. On the one hand, we find that all existing threat models areindeed applicable. On the other hand, there are significant mismatches:research is often too generous with the attacker, assuming access toinformation not frequently available in real-world settings. Our paper is thusa call for action to study more practical threat models in artificialintelligence security.

Towards more Practical Threat Models in Artificial Intelligence Security