PitchBook analyst Brendan Burke presented an in-depth discussion with the CEO of HydroX AI.

Security, privacy,and copyright issues are top of enterprise user minds when considering building applications around large language models (LLMs). Urgently responding to these concerns, hyperscalers and startups have made secure deployments a cornerstone of their product offerings. In this Tech Talks webinar, CEO Zhuo Li of HydroX AI joined PitchBook Senior Analyst Brendan Burke to discuss test results and research on enhancing LLM safety. HydroX AI has developed some of the most extensive testing suites for LLMs in security, privacy, safety and integrity, partnering with the leading industry association AI Alliance.

To watch the recording please fill out the form.

Key Topics

• How open-source models compare to proprietary models in safety and security
  
  • HydroX’s latest learnings on cross-gen comparison of Llama models
  • Insights from the recent Llama 3.1 release
• How enterprises should think about mitigating the safety and security risks (e.g. model selection, model guardrails)
• Progress on HydroX AI’s latest research on evaluation of LLM for domain-specific safety and alignment, in collaboration with IBM, Meta, and the AI Alliance
• How individual users can learn more about red-teaming and model safety

Related PitchBook Reports

PitchBook Analyst Note: High-stakes foundation model horse race out of the gates
Q2 2024 Information security report
PitchBook Analyst Note: 2024 RSA conference coaxes information security into the LLM age