AI's Achilles' Heel: Startups Emerge to Secure the Generative AI Stack
The increasing adoption of generative AI models has brought new security challenges. Startups are emerging to address these challenges by providing solutions in three categories: governance, observability, and security.
Summary
- Generative AI models are increasingly vulnerable to cyberattacks. These attacks can target the models themselves (e.g., model theft) or exploit them to generate malicious content (e.g., phishing emails).
- There are three main categories of security solutions for generative AI: governance, observability, and security. Governance solutions help organizations understand and manage their AI usage. Observability tools enable organizations to monitor the behavior of their AI models. Security solutions protect models from attacks and misuse.
- Some of the most promising startups in the generative AI security space include Robust Intelligence, Lakera, Prompt Security, Private AI, Nightfall, Hiddenlayer, Lasso Security, DynamoML, FedML, Tonic, Gretel, Private AI, Kobalt Labs, Protect AI, and Giskard.
- Menlo Ventures is actively investing in generative AI security startups. They are looking for teams with deep expertise in AI infrastructure, governance, and security.