BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//pretalx//pretalx.com//hack-lu-2025//talk//YXLGPP
BEGIN:VTIMEZONE
TZID:CET
BEGIN:STANDARD
DTSTART:20001029T040000
RRULE:FREQ=YEARLY;BYDAY=-1SU;BYMONTH=10
TZNAME:CET
TZOFFSETFROM:+0200
TZOFFSETTO:+0100
END:STANDARD
BEGIN:DAYLIGHT
DTSTART:20000326T030000
RRULE:FREQ=YEARLY;BYDAY=-1SU;BYMONTH=3
TZNAME:CEST
TZOFFSETFROM:+0100
TZOFFSETTO:+0200
END:DAYLIGHT
END:VTIMEZONE
BEGIN:VEVENT
UID:pretalx-hack-lu-2025-YXLGPP@pretalx.com
DTSTART;TZID=CET:20251022T114500
DTEND;TZID=CET:20251022T121500
DESCRIPTION:This session dissects a real-world case study where an actor we
 aponized automation flaws in Meta’s LLM-based compliance system to hijac
 k high-value accounts via orchestrated botnet abuse\, prompt injection\, a
 nd linguistic manipulation. The attacker exploited vulnerabilities in the 
 very safeguards designed to protect users\, triggering account suspension 
 and negotiating “restoration” through AI-manipulated support flows.\n\
 nThis case is not an isolated incident—it is a signal of broader systemi
 c risks that emerge when generative models and automation pipelines are in
 tegrated without robust adversarial testing. Beyond the technical compromi
 se\, the attack leveraged prompt engineering as social engineering\, revea
 ling the cognitive blind spots of model-aligned trust systems.\n\nIn respo
 nse\, I introduce foundational forensic linguistic techniques and NLP-base
 d detection methods for identifying AI-generated text in compromised commu
 nications. By combining stylometry\, perplexity analysis\, and syntax anom
 aly detection in Python\, we illuminate detection opportunities hidden in 
 prompts and narrative structure. With few more tips from cloud security ar
 ea to protect the LLM deployments.\nThe talk closes with a reflection on t
 he ethical tensions in detecting synthetic media.\n\nThis talk will blend 
 live demonstration\, code walkthroughs\, and operational insights from an 
 investigation that didn’t just uncover an exploit—but a philosophy of 
 misuse.
DTSTAMP:20260714T165034Z
LOCATION:Europe
SUMMARY:Smack my LLM up! - Jindrich Karasek
URL:https://pretalx.com/hack-lu-2025/talk/YXLGPP/
END:VEVENT
END:VCALENDAR