DeepKeep

DeepKeep is an Israeli startup founded in 2021 by Rony Ohayon that provides AI-native security and trustworthiness for artificial intelligence systems. The company has secured $10 million in a seed funding round led by Awz Ventures. The platform is designed to safeguard the entire AI lifecycle, from the research and development phase through to deployment and monitoring. The company was established by Rony Ohayon, who serves as the CEO. Ohayon has approximately 20 years of experience in the high-tech industry, holds a Ph.D. in Communication Systems Engineering, and has founded other technology companies, including DriveU and LiveU. The management team also includes Yossi Altevet as CTO, Gad Sosa as VP of Research & Development, and Ofer Rotberg as VP of Product. DeepKeep operates in the Trust, Risk, and Security Management (TRiSM) market. Its business model is B2B, offering its software to large enterprises in sectors such as finance, security, automotive, and AI computing. The company employs a go-to-market strategy that includes direct sales and channel partnerships, offering its platform as both a cloud-based SaaS and an on-premises deployment to accommodate various client needs, including air-gapped environments. The core of DeepKeep's offering is its AI-Native security platform, which uses generative AI to secure AI, generative AI, and large language models (LLMs). The platform is model-agnostic, meaning it can protect a variety of AI models, including computer vision, LLMs, and tabular data models. Key features include continuous risk assessment through penetration testing, automated prevention and hardening of models before deployment, and real-time detection of attacks. The product portfolio consists of modules like an AI Firewall for real-time alerts, Automated AI Red Teaming for robustness evaluation, Model Scanning to secure the AI supply chain, and AI Lens for monitoring employee interactions with AI systems. The platform addresses a range of threats including data leakage, prompt injection, model hallucinations, bias, and toxic responses. Keywords: AI security, Trust Risk and Security Management, TRiSM, large language model security, generative AI protection, AI lifecycle security, model-agnostic security, computer vision security, AI firewall, automated red teaming, AI risk assessment, machine learning security, data leakage prevention, adversarial attack mitigation, AI model governance, Rony Ohayon, AI trustworthiness, secure AI deployment, enterprise AI security, cybersecurity for AI