BEGIN:VCALENDAR
VERSION:2.0
X-WR-CALNAME:BrightTALK Event
PRODID:-//BrightTALK//NONSGML BrightTALK Event Calendar//EN
CALSCALE:GREGORIAN
METHOD:REQUEST
BEGIN:VEVENT
UID:https://www.brighttalk.com/webcast/18657/656685
DTSTAMP:20251104T054238Z
ORGANIZER:MAILTO:noreply@brighttalk.com
LOCATION:BrightTALK
URL:https://www.brighttalk.com/webcast/18657/656685?utm_campaign=commu
 nication_reminder_24hr_registrants&utm_medium=calendar&utm_source=brig
 httalk-transact
DTSTART:20260603T050000Z
DTEND:20260603T055936Z
SUMMARY:Live webcast: Unreliable AI: Addressing the Cybersecurity Risk
 s of LLM-Based System
DESCRIPTION:Click here to attend: https://www.brighttalk.com/webcast/1
 8657/656685?utm_campaign=communication_reminder_24hr_registrants&utm_m
 edium=calendar&utm_source=brighttalk-transact\n\nPresenter: Vladislav 
 Tushkanov - Research Development Group Manager | Machine Learning Tech
 nology |AI Research | Kaspersky\n\nAs AI systems based on large langua
 ge models become more powerful and widespread, more cybersecurity chal
 lenges emerge. From jailbreaks to indirect prompt injections, they are
  vulnerable to a wide array of LLM-specific threats. However, all thes
 e problems can be boiled down to one single issue: LLMs are probabilis
 tic algorithms that are inherently unreliable, just like any other ML 
 system. Does it mean they cannot be useful? Absolutely not! In this ta
 lk, we will discuss the challenges of secure and reliable LLM-based ap
 plications and the way to make them safer and more aligned with your b
 usiness goals.
SEQUENCE:1780375572
END:VEVENT
END:VCALENDAR