Webcast
Multimodal AI:
The Possibility and Peril
Webcast: Multimodal AI – The Possibility and Peril
Get the 30-minute recording and slides to learn:
- Why multimodal AI systems are harder to safeguard,
- Details on why 2 Mistral models were found to be 60X more likely to generate CSEM responses and 18-40X more likely to generate CBRN responses compared to other models
- Real examples of how image+text jailbreaks are bypassing filters and revealing safety risks
- Practical steps to secure multimodal AI (including a short product demo)
Whether you're building with AI or evaluating your risk posture, this is a must-watch for staying ahead of the AI threat landscape.
.svg){kind=small}
