The agentic generation of security and cloud operations starts in the Dojo

Yes. Dojo AI leverages foundation models securely hosted through Amazon Bedrock.

Yes. Capabilities like Mobot and the SOC Analyst Agent process customer telemetry to perform investigations, run queries, and generate findings. All processing occurs securely within your platform context, and customer data is never used to train generalized AI models. Administrators can turn off Dojo AI capabilities at any time through Feature Management or by submitting a support ticket.

Customers that previously opted out of Sumo Logic AI capabilities will not get access to these or future AI capabilities until they explicitly opt back in.

Sumo Logic AI capabilities follow strict legal, compliance, and security standards to ensure data minimization and fit-for-purpose processing.

• Customer data is never used to train AI models, shared externally, or used to improve global models.
• Data remains within the customer’s environment and is processed only to deliver results back to that customer.
• Sumo Logic applies strong safeguards and filtering to ensure sensitive data is handled securely and appropriately at all times.

Customers can opt out of capabilities that process customer data, including the SOC Analyst Agent and Mobot, at any time from Feature Management or by submitting a support ticket.

No. Customer data is never used to train AI models.

All Sumo Logic AI capabilities are designed to serve customer-specific outcomes within their own environment. Mobot uses a large language model (LLM) via Amazon Bedrock, which processes data securely and does not retain or use customer information for training or other external purposes.

Traditional ML features, such as AI-driven alerts, generate models specific to each customer’s environment and are never shared or made public.

For more information, see trust.sumologic.com.

No additional third parties have any access. Dojo AI leverages foundation models securely hosted through Amazon Bedrock. When customer data is processed using Amazon Bedrock:

• Customer inputs and outputs are treated as Customer Content under AWS terms.
• AWS does not use Customer Content to train models or improve Amazon Bedrock.
• AWS may access Customer Content only as necessary to provide the service or comply with law.
• Third-party model providers do not have access to customer inputs or outputs.
• Customer inputs and outputs are not shared with model providers and are not used to train external models.

Customer data processed through Dojo AI remains within Sumo Logic’s secure environment and is used only to deliver results for that customer. It is not used shared with model providers.

Dojo AI and classical ML features store data only temporarily to optimize performance:

• AI-driven alerts use a rolling 60-day data window, retraining weekly and expiring the oldest data automatically.
• Mobot may temporarily retain conversation history in a rolling window to improve conversational context and response accuracy.

All stored data follows Sumo Logic’s data retention and deletion policies, ensuring customer information is never retained longer than necessary.

Sumo Logic is currently reviewing AI compliance within a rapidly evolving framework, in particular ISO 42001, designed to help organizations implement AI responsibly.

Sumo Logic AI capabilities operate within our existing industry-recognized security and compliance framework, including FedRAMP Moderate, SOC 2 Type 2, HIPAA, PCI DSS 4.0.1, and ISO 27001:2022. These attestations govern the confidentiality, integrity, and protection of customer data.

Availability of specific AI capabilities may vary by deployment region (including FED) based on compliance boundary requirements.

The current GA versions of Mobot (including Query Agent and Knowledge Agent) and Summary Agent are available in the FED deployment.

The SOC Analyst Agent and certain newer Dojo AI capabilities are not currently available in FED. These capabilities depend on underlying model configurations that do not yet meet the requirements of our FED compliance boundary.

We are actively evaluating future availability of these capabilities in FED as underlying model support and compliance requirements evolve.

The generative AI model is licensed and securely hosted via Amazon Bedrock, meaning it is not directly accessible by Sumo Logic, customers, or third parties.

All new AI capabilities and features undergo comprehensive legal, compliance, and application security reviews before release to ensure data protection, privacy, and regulatory alignment.

Recurring reviews are also conducted with every major update, particularly when a capability introduces new analytics or processes previously unused data types, to maintain ongoing trust and compliance.

AI features are on by default. We offer two methods for opting out.

In-product self opt-out: Beginning August 2026, platform administrators can opt out of all AI features directly within the product settings under Feature Management. Disabling AI at the admin level automatically exempts your tenant from any future AI feature rollouts.

Via your account team or support: You can opt out of specific AI features or all AI features by contacting your Sumo Logic account team or opening a support ticket. If you choose to opt out of all AI features, your tenant will automatically be exempted from future AI feature releases.

For a complete overview of everything included in Sumo Logic’s AI portfolio, click here.