AI Safety, Security & Red-Teaming
11 companies
0 of 0 categories open
Relationship:
Loading the map…
On this page
Organisations working on making models safer to deploy, split across two quite different activities. One is commercial security products defending deployed applications against attack. The other is research organisations evaluating model capabilities and risks.
What to compare
Decide which of those two you actually want, first. For security products, compare coverage of prompt injection, data leakage and model theft, whether protection happens at runtime or before deployment, and how it fits the security tooling already in place. For evaluation work, independence and methodology outweigh product features.
11 companies in this category
Adversa AIAI red-teaming and security research firm.
Apollo ResearchAI safety research org focused on detecting deceptive model behavior.- Calypso AIAI security platform for validating and securing model deployments.
CraniumAI security and risk-management platform for enterprises.
Credo AIAI governance platform for responsible AI risk management.
HiddenLayerSecurity platform defending ML models from adversarial attacks.
LakeraAI security platform protecting LLM applications from prompt attacks.
METRNonprofit evaluating frontier models for dangerous autonomous capabilities.
Protect AISecurity platform for scanning and securing the ML/AI supply chain.
Redwood ResearchAI safety research organization focused on alignment techniques.
Robust IntelligenceAI model risk and security platform (acquired by Cisco).