Security
Security information by product
SternBench Search
What the product is
SternBench Search is the legal research database in the SternBench product family. It is used to search, filter, and read offshore judgments across the covered jurisdictions.
Typical data involved
SternBench Search is built around public-source judicial material. Typical data includes public judgments, public case metadata, search queries, and limited auth, rate-limit, and aggregate usage metadata needed to operate and protect the service. SternBench does not keep its own server-side search history or query-keyed search cache within the application. Search queries are processed by Google Vertex AI Search, and provider-side logging or analytics may apply. Search is designed for research across public-source judgments rather than for submitting firm-confidential, privileged, personal, or other sensitive matter information, and it does not depend on users uploading firm documents as part of ordinary use.
Access model
SternBench Search is available for direct use. Firm access options are available where broader institutional access is required.
Hosting and processing
SternBench Search uses Google Cloud services for search, storage, and application data. Search retrieval is delivered through Google Vertex AI Search. The web frontend is delivered through Vercel.
AI and model training
Search uses AI-assisted retrieval to improve ranking and discovery across the case law corpus. According to Google Cloud's Vertex AI Search data-governance documentation, customer data used in Vertex AI Search is not used to train foundation models. In the current Search application flow, search requests are processed through Google Vertex AI Search and are not routed by the application to OpenAI or Anthropic.
Access controls
Production access is restricted. Administrative access is limited to authorized internal personnel and used only for operating, maintaining, and securing the service.
Core safeguards
Data is encrypted in transit and at rest. The service is protected through access, rate-limit, and operational controls. Because Search is built on public-source judicial material and does not rely on firm document upload, its data profile is narrower than document-centric legal AI products.
Procurement and diligence
Additional security and legal materials can be provided for firm review on request, including privacy, data handling, and procurement follow-up materials.
SternBench Research
What the product is
SternBench Research is the AI research product in the SternBench family. It allows users to ask research questions across the full SternBench case law database and move from an answer to the underlying authorities.
Typical data involved
Typical data includes public judgments, structured case law data, user prompts, generated research outputs, and related research session context.
Access model
Research is not positioned as a broad public self-serve product. It is introduced through request-based access and controlled onboarding.
Hosting and processing
SternBench Research is hosted on Google Cloud infrastructure. AI responses are delivered through Google Vertex AI and Gemini on top of the SternBench case law database.
AI and model training
User prompts and generated outputs are not used to train underlying AI models. Research is designed to operate on the SternBench database so that the research workflow remains tied to the underlying authorities.
Access controls
Access to Research is controlled at the product level. Production access is restricted, and administrative access is limited to authorized internal personnel.
Core safeguards
Data is encrypted in transit and at rest. The service is protected through standard access, monitoring, and operational controls. Research is designed so users can work from a research question back to the relevant authorities in the database rather than treating the AI layer as separate from the source material.
Procurement and diligence
Additional security and legal materials can be provided for firms reviewing SternBench Research, including product-specific procurement follow-up where needed.
SternBench Assistant
What the product is
SternBench Assistant is the advanced legal AI workspace in the SternBench product family. It is in the same general product category as Harvey and Legora: a system that can read documents, draft and revise written work, and work with the user across a matter while also having access to the SternBench case law database.
Typical data involved
Typical data includes public judgments, user prompts, uploaded documents, draft work product, matter materials, and workspace context.
Access model
Assistant is not a general public self-serve product. It is introduced through request-based access and controlled onboarding.
Hosting and processing
SternBench Assistant is hosted on Google Cloud infrastructure and is designed to combine document workflow with the SternBench case law database inside one controlled environment.
AI and model training
Uploaded materials, user prompts, and generated work product are not used to train underlying AI models. Assistant is designed so document and research work can occur within one environment rather than being split between a general AI tool and a separate research database.
Access controls
Assistant has the strictest access model in the SternBench family. Access is controlled through request-based onboarding. Production access is restricted, and support or administrative access is tightly limited and handled through controlled internal processes.
Core safeguards
Data is encrypted in transit and at rest. The service is protected through standard access, monitoring, and operational controls appropriate to a document-workflow product. Because Assistant can involve uploaded working materials and draft matter documents, it is reviewed separately from Search and Research in firm procurement processes.
Procurement and diligence
Firms evaluating SternBench Assistant should expect a product-specific diligence conversation. Additional security, privacy, and procurement materials are available on request.
Security Questions
Short answers to common first-pass diligence questions.
Because SternBench Search, SternBench Research, and SternBench Assistant handle different types of data and support different workflows. Firms should review the section for the specific product they are evaluating.
SternBench products use Google Vertex AI and Gemini for AI features, with product-specific differences in how those capabilities are applied.
No. User prompts, uploaded materials, and generated outputs are not used to train underlying AI models.
No. SternBench Research and SternBench Assistant are introduced through request-based access and controlled onboarding.
Yes. Additional security, privacy, and procurement materials are available for firms conducting a formal review.
Use the contact route provided below for procurement, security, or legal diligence follow-up.
Security review and diligence requests
SternBench can provide additional security, privacy, and procurement materials for firms conducting a formal review of Search, Research, or Assistant.