Data retention policy
Subscriber Content is retained for the duration of the agreement, in line with the retention instructions given by the customer. After termination or expiry, Legora assists with retrieval of Subscriber Content for 30 days and then securely deletes all remaining Subscriber Content using industry-standard data destruction practices, unless applicable law requires retention. Written confirmation of deletion is provided on request. Critical security logs are retained for at least 12 months.
Data archiving and removal policy
At the customer's choice and instruction, Legora deletes or returns all personal data when the data processing agreement ends, and deletes Subscriber Content on written request. If Legora is legally required to keep archival copies (for example for tax purposes), it informs the customer, uses the copies only to meet that obligation, and keeps them under the same confidentiality and security obligations. Production backups are taken every 4 hours and are encrypted in transit and at rest.
Data storage policy
Customer data is stored in Microsoft Azure in the customer's assigned region (EU, US, Asia-Pacific or Australia). Files that are indexed for search are held by our search infrastructure provider, as listed on our sub-processor page. Data is encrypted with AES-256 (or equivalent) at rest and TLS 1.2 or higher in transit. Customer environments are logically separated, and development, test, staging and production environments are kept apart.
App/service has sub-processors
yes
Guidelines for sub-processors
App/service uses large language models (LLM)
yes
LLM model(s) used
Anthropic Claude (Sonnet and Opus) and OpenAI GPT models, hosted on AWS Bedrock, Google Vertex AI, Azure OpenAI and OpenAI depending on the customer's region. Availability varies by region
LLM retention settings
Legora’s customer terms prohibit sub-processors, other than cloud storage providers, from retaining Subscriber Content or logging it for human review. Model providers are contracted under zero-data-retention terms.
LLM data tenancy policy
Customer environments are logically separated. Legora does not use a customer's Confidential Information, including Subscriber Content, to train or fine-tune AI models, and does not allow its subcontractors to do so. Usage data Legora collects is limited
LLM data residency policy
Model processing takes place with the sub-processors and in the locations listed for the customer's region on our sub-processor page: EU/EEA for EU customers and the US for US customers. For Asia-Pacific and Australian customers, model processing may take