Repello AI, an AI security firm based in San Francisco and Bengaluru, has raised US$1.2 million in seed funding.

The funding round included participation from Venture Highway, pi Ventures, Entrepreneur First, and angel investors such as Charles Songhurst, Vivek Raghavan, and Satya Vyas.

The company plans to use the funds to improve its AI security products, Artemis and Repello Guard.

Founded in 2024 by IIT Roorkee alumni Aryaman Behera and Naman Mishra, Repello AI addresses risks related to generative AI, including data breaches, compliance failures, and unsafe outputs.

Its clients include companies like Groww and PhysicsWallah.

🔗 Source: Repello

🧠 Food for thought

1️⃣ Red teaming emerges as critical infrastructure amid rising AI security concerns

Repello AI’s funding comes as AI security risks are projected to surge by 50% in 2024, with organizations facing daily AI-driven attacks 1.

This investment aligns with a broader market trend, as the AI security sector is expected to grow to $60.24 billion by 2029, signaling increasing enterprise awareness of AI-specific vulnerabilities 1.

Established cybersecurity players are also recognizing this opportunity, with CrowdStrike recently launching dedicated AI Red Team Services to identify vulnerabilities in AI systems, including prompt injection attacks and data poisoning 2.

The growing ecosystem of specialized tools, such as NVIDIA’s Garak vulnerability scanner and Microsoft’s PyRIT for AI security assessment, demonstrates how red teaming is becoming a critical component for AI deployment 3.

Repello’s approach resembles other specialized platforms like Mindgard, which automates AI security testing throughout the development lifecycle, indicating a market shift toward continuous rather than periodic security validation 4.

2️⃣ AI security evolves beyond traditional cybersecurity frameworks

The emergence of companies like Repello highlights a significant shift from traditional cybersecurity approaches to AI-specific defenses addressing novel threat vectors 5.

While early machine learning applications in cybersecurity focused on anomaly detection and spam filtering, today’s AI security tools must address unique threats like model hallucinations, adversarial prompts, and training data leakage 6.