BEGIN:VCALENDAR VERSION:2.0 PRODID:-//Sabre//Sabre VObject 4.5.8//EN CALSCALE:GREGORIAN BEGIN:VTIMEZONE TZID:Europe/Zurich X-LIC-LOCATION:Europe/Zurich TZURL:http://tzurl.org/zoneinfo/Europe/Zurich BEGIN:DAYLIGHT TZOFFSETFROM:+0100 TZOFFSETTO:+0200 TZNAME:CEST DTSTART:19810329T020000 RRULE:FREQ=YEARLY;BYMONTH=3;BYDAY=-1SU END:DAYLIGHT BEGIN:STANDARD TZOFFSETFROM:+0200 TZOFFSETTO:+0100 TZNAME:CET DTSTART:19961027T030000 RRULE:FREQ=YEARLY;BYMONTH=10;BYDAY=-1SU END:STANDARD END:VTIMEZONE BEGIN:VEVENT UID:news1929@dmi.unibas.ch DTSTAMP;TZID=Europe/Zurich:20260206T174818 DTSTART;TZID=Europe/Zurich:20260127T161500 SUMMARY: The Practice-Research Gap in AI Threat Modeling DESCRIPTION:Abstract: Cybersecurity ensures the trustworthy and reliable fu nctioning of digital systems. Currently\, companies spend about 10% of the ir IT budget on cybersecurity. Thus\, security and threat modelling become increasingly relevant also for technologies in artificial intelligence. H owever\, existing AI threat models have faced criticism regarding their pr acticality. Common issues include\, but are not limited to\, unrealistic a ssumptions\, a focus on isolated models rather than full AI pipelines\, an d perturbation techniques that lack real-world applicability.\\r\\nTo addr ess these gaps\, one approach is to measure how AI systems are deployed in practice and assess their exposure to known attack vectors. An orthogonal strategy involves collecting empirical data on real-world AI incidents th rough systematic reporting. Lastly\, AI applications can be threat modelle d from the earliest stages. To this end\, we examine a real-world applicat ion of AI\, the electric grid\, and review caveats and implications for AI security.\\r\\nDr. Kathrin Grosse [https://research.ibm.com/people/kathri n-grosse] X-ALT-DESC:

Abstract: Cybersecurity ensures the trustworthy and reliable functioning of digital systems. Currently\, companies spend about 10% of t heir IT budget on cybersecurity. Thus\, security and threat modelling beco me increasingly relevant also for technologies in artificial intelligence. However\, existing AI threat models have faced criticism regarding their practicality. Common issues include\, but are not limited to\, unrealistic assumptions\, a focus on isolated models rather than full AI pipelines\, and perturbation techniques that lack real-world applicability.

\n

To address these gaps\, one approach is to measure how AI systems are deploy ed in practice and assess their exposure to known attack vectors. An ortho gonal strategy involves collecting empirical data on real-world AI inciden ts through systematic reporting. Lastly\, AI applications can be threat mo delled from the earliest stages. To this end\, we examine a real-world app lication of AI\, the electric grid\, and review caveats and implications f or AI security.

\n

Dr. Kathrin Grosse

END:VEVENT END:VCALENDAR