HydroX AI secures $4 million in funding to enhance AI safety and security for large language models (LLMs). The introduction of the EPASS Evaluation Platform aims to provide detailed assessments and management tools for AI models, addressing critical concerns in the AI TRiSM sector.
HydroX AI Secures $4 Million Funding for AI Security Platform
San Jose, Calif. – June 25, 2024 – HydroX AI, an emerging player in artificial intelligence (AI) safety and security, has successfully closed a $4 million angel funding round. The investment was backed by Vitalbridge Capital, Atom Capital, and Qi Lu, a prominent figure in the AI industry and former COO of Baidu. The funding round will support HydroX AI’s efforts to innovate and validate its market-focused safety platform for large language models (LLMs).
As the global demand for LLMs continues to rise, so too does the concern over their safety and security. The LLM market is projected to hit $260 billion by 2030, emphasising the need for stringent security measures. HydroX AI notes that LLMs can become vulnerable to compromises within just three months of their release, highlighting the necessity for continuous updates and improvements in security protocols.
HydroX AI has positioned itself within the AI Trust, Risk, and Security Management (AI TRiSM) sector to address these risks. According to Ben, a partner at Vitalbridge Capital, “The integration of AI into society brings numerous challenges including privacy, security, ethics, and psychological impacts. Security will be essential not just for protection but as a fundamental part of AI’s growth and trustworthiness.”
Introduction of EPASS
Alongside the funding announcement, HydroX AI also launched EPASS (Evaluation Platform for AI Safety and Security). This platform aims to provide detailed assessments and management tools for AI models, offering critical insights into areas such as safety, security, privacy, and integrity. The EPASS platform is expected to become an essential tool in navigating the ethical, social, and existential risks associated with AI technologies.
The platform offers several key features:
– Evaluation: It assesses AI models using advanced techniques to identify vulnerabilities, ensuring models meet high safety and integrity standards. Detailed reports provide actionable insights for performance and reliability enhancements.
– Management: A user-friendly dashboard allows for seamless model management and comparison, integrating with existing workflows for real-time monitoring and updates.
– Leaderboard: An extensive ranking system benchmarks AI models across more than 30 safety and security categories, giving users valuable reference points for improvement.
This comprehensive tool is designed to enable stakeholders across varied sectors to manage and compare AI models without the need for extensive in-house evaluation infrastructure. Customers can select from essential features included in standard accounts, such as one daily model evaluation and limited attack methods, to more advanced customisable options like unlimited evaluations, tailored category bundles, and dedicated GPU resources.
Early Development and Engagement
Currently in its early stages, HydroX AI is engaging with major LLM companies and has already reported significant improvements in AI model performance and security, thanks to EPASS. This has garnered endorsements from several leading LLM companies, reinforcing HydroX AI’s pioneering stance and first-mover advantage in the critical field of AI safety and security.
HydroX AI was founded in 2023 by Zhuo Li, former head of the Privacy and Data Protection Office at Bytedance. Li boasts a comprehensive background in privacy and security from his time at Meta and LinkedIn. The team also includes Yuji Kosuga, Head of Engineering, who brings a wealth of experience from LinkedIn, Google, and Meta, further bolstering the company’s dedication to advancing AI safety and security.
Future Outlook
As AI continues to integrate into various aspects of society and industry, the importance of robust safety and security frameworks cannot be overstated. With its recent financial backing and the launch of the EPASS platform, HydroX AI is well-positioned to address the growing complexities and risks associated with LLMs. The company’s efforts signal a significant step towards safer and more reliable AI systems in the future.
