BEGIN:VCALENDAR VERSION:2.0 PRODID:IEEE vTools.Events//EN CALSCALE:GREGORIAN BEGIN:VTIMEZONE TZID:America/Los_Angeles BEGIN:DAYLIGHT DTSTART:20250309T030000 TZOFFSETFROM:-0800 TZOFFSETTO:-0700 RRULE:FREQ=YEARLY;BYDAY=2SU;BYMONTH=3 TZNAME:PDT END:DAYLIGHT BEGIN:STANDARD DTSTART:20251102T010000 TZOFFSETFROM:-0700 TZOFFSETTO:-0800 RRULE:FREQ=YEARLY;BYDAY=1SU;BYMONTH=11 TZNAME:PST END:STANDARD END:VTIMEZONE BEGIN:VEVENT DTSTAMP:20260124T050207Z UID:6BD60B89-88D4-4FAE-B102-0A13FEE72783 DTSTART;TZID=America/Los_Angeles:20250624T170000 DTEND;TZID=America/Los_Angeles:20250624T190000 DESCRIPTION:Generic Large Language Models (GLLMs) are continually being rel eased with increased size and capabilities\, enhancing the capabilities of these tools as universal problem solvers. While the reliability of GLLMs' responses is questionable in many situations\, these models are often aug mented or retrofitted with external resources for various applications\, i ncluding cybersecurity.\n\nThe talk will discuss major security concerns o f these pre-trained models: first\, GLLMs are prone to adversarial manipul ation\, such as model poisoning\, reverse engineering\, and side-channel c yberattacks. Second\, the security issues related to LLM-generated codes u sing open-source libraries/codelets for software development can involve s oftware supply chain attacks. These may result in information disclosure\, access to restricted resources\, privilege escalation\, and complete syst em takeover.\n\nThis talk will also cover the benefits and risks of using GLLMs in cybersecurity\, particularly in malware detection\, log analysis\ , intrusion detection\, etc. I will highlight the need for diverse AI appr oaches (non-LLM-based smaller models) trained with application-specific cu rated data\, fine-tuned for well-tested security functionalities in identi fying and mitigating emerging cyber threats\, including zero-day attacks.\ n\nNote:\n\n- You will require a Zoom account (free to obtain) to join the meeting. This requirement is to avoid Zoom bombing. Please sign in using the email address tied to your Zoom account\, not necessarily the one you used to register for the event. Register here: https://sjsu.zoom.us/meetin g/register/2XuaGc9ISoCWOu1dt6ANog\n- By registering for this event\, you a gree that IEEE and the organizers are not liable to you for any loss\, dam age\, injury\, or any incidental\, indirect\, special\, consequential\, or economic loss or damage (including loss of opportunity\, exemplary or pun itive damages). The event will be recorded and will be made available for public viewing.\n\nCo-sponsored by: Vishnu S. Pendyala\, SJSU\n\nSpeaker(s ): Dr. Vishnu S. Pendyala\, Prof. Dipankar Dasgupta\, IEEE Fellow\, NAI Fe llow\, AIIA Fellow\n\nVirtual: https://events.vtools.ieee.org/m/489327 LOCATION:Virtual: https://events.vtools.ieee.org/m/489327 ORGANIZER:pendyala@ieee.org SEQUENCE:32 SUMMARY:Generic LLMs in Cybersecurity URL;VALUE=URI:https://events.vtools.ieee.org/m/489327 X-ALT-DESC:Description: <br /><p>Generic Large Language Models (GLLMs) are continually being released with increased size and capabilities\, enhancin g the capabilities of these tools as universal problem solvers. &nbsp\;Whi le the reliability of GLLMs' responses is questionable in many situations\ , these models are often augmented or retrofitted with external resources for various applications\, including cybersecurity.</p>\n<p><br>The talk w ill discuss major security concerns of these pre-trained models: first\, G LLMs are prone to adversarial manipulation\, such as model poisoning\, rev erse engineering\, and side-channel cyberattacks. Second\, the security is sues related to LLM-generated codes using open-source libraries/codelets f or software development can involve software supply chain attacks. These m ay result in information disclosure\, access to restricted resources\, pri vilege escalation\, and complete system takeover.</p>\n<p><br>This talk wi ll also cover the benefits and risks of using GLLMs in cybersecurity\, par ticularly in malware detection\, log analysis\, intrusion detection\, etc. I will highlight the need for diverse AI approaches (non-LLM-based smalle r models) trained with application-specific curated data\, fine-tuned for well-tested security functionalities in identifying and mitigating emergin g cyber threats\, including zero-day attacks.</p>\n<p>Note:</p>\n<ul type= "disc">\n<li>You will require a Zoom account (free to obtain) to join the meeting. This requirement is to avoid Zoom bombing. Please sign in using t he email address tied to your Zoom account\, not necessarily the one you u sed to register for the event. Register here: <a href="https://sjsu.zoom.u s/meeting/register/2XuaGc9ISoCWOu1dt6ANog">https://sjsu.zoom.us/meeting/re gister/2XuaGc9ISoCWOu1dt6ANog</a>&nbsp\;</li>\n<li>By registering for this event\, you agree that IEEE and the organizers are not liable to you for any loss\, damage\, injury\, or any incidental\, indirect\, special\, cons equential\, or economic loss or damage (including loss of opportunity\, ex emplary or punitive damages). The event will be recorded and will be made available for public viewing.</li>\n</ul> END:VEVENT END:VCALENDAR