BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//pretalx//pretalx.com//orangecon-2026//speaker//MEYDL8
BEGIN:VEVENT
UID:pretalx-orangecon-2026-XGZYKB@pretalx.com
DTSTART:20260604T145000Z
DTEND:20260604T152000Z
DESCRIPTION:Many organizations are developing LLM‑based applications to i
 mprove productivity\, supported by the growing number of platforms that si
 mplify their creation. However\, integrating LLMs into applications introd
 uces new security risks\, as adversaries can exploit models through natura
 l‑language–based attacks such as prompt injections and jailbreaks. Suc
 cessful attacks can lead to sensitive data leakage\, reputational harm\, o
 r deeper compromise of internal digital environments. \n\nThese risks high
 light the need for structured\, repeatable\, and context‑aware security 
 testing for LLM‑enabled applications. Therefore\, we would like to prese
 nt ProViLE: a systematic approach and supporting open‑source tool for pr
 ompt‑based security testing of LLM‑enabled applications. ProViLE empha
 sizes that effective tests are highly dependent on the context of the appl
 ication. The approach guides practitioners through four key steps: (1) def
 ining potential attack objectives\, (2) identifying relevant attack techni
 ques\, (3) formulating corresponding attack prompts\, and (4) evaluating t
 he LLM application’s responses to the attack prompts.\n\nThe ProViLE too
 l automates the final two steps by using LLMs to (3) generate attack promp
 ts from objectives and techniques\, and (4) evaluate whether a response co
 nstitutes a successful attack based on the objective and a scoring rubric.
  This enables scalable and consistent testing across diverse application c
 ontexts. The result is a structured overview of the security posture of an
  LLM‑based application across custom security considerations.\n\nProViLE
  aims to facilitate the penetration‑testing workflow for LLM application
 s\, but can also be used by development teams to conduct initial baseline 
 assessments before deployment. By open‑sourcing our work\, we hope to su
 pport the broader development of secure LLM‑based systems.
DTSTAMP:20260525T192652Z
LOCATION:Track 2
SUMMARY:How to Prompt for Vulnerabilities in LLM-based applications with Ex
 tensions\, the ProViLE approach. - Rajeck Massa
URL:https://pretalx.com/orangecon-2026/talk/XGZYKB/
END:VEVENT
END:VCALENDAR