Navigating The Storm: Understanding the Potential Threat Landscape of Cloud GenAI Services

The MITRE ATT&CK Framework v.16 was released in October 2024 with over 19 new techniques. This blog discusses two new Cloud Matrix sub-techniques: Modify Authentication Process: Conditional Access Policies (T1556) and Resource Hijacking: Cloud

With GenAI adoption rising significantly in various sectors over the past years, the threat landscape associated with GenAI will also evolve rapidly. Learn more about what security challenges can potentially threaten the cloud GenAI workloads.

LLMJacking attacks are increasingly targeting GenAI workloads on Amazon Bedrock. Cybercriminals illegally gain access to Large Language Models during LLMJacking attacks and resell this access via chatbots for profit.

We are excited to release powerful features to the Mitigant Security Platform: Attack Emulation for AWS GenAI services, Mitigant API, and support for AWS Organizations. These features empower organizations to use GenAI cloud services securely.

This article shows how to leverage Threat-Informed Defense to effectively tackle cloud threat actors, e.g., Scattered Spider. Practical attacks & appropriate defense are demonstrated using Mitigant Cloud Attack Emulation & the Sekoia SOC Platform.

Amazon Bedrock's Achilles heel is the adoption of S3 as a RAG data source, leading to several GenAI attack vectors, including data poisoning, denial of service & S3 ransomware. This article delves into these attack vectors, detection, and mitigation.