Bedrock Data expands AI security to Confluence

Bedrock Data has added native support for Atlassian Confluence, expanding its data security posture management coverage into collaboration content that increasingly feeds AI systems.

The company said organisations store intellectual property, customer records and internal plans in Confluence. Teams also connect that material to retrieval-augmented generation systems and agent workflows. Bedrock Data said security teams often lack visibility into what information AI models can access and what they might return during inference.

Bedrock Data positions the integration as a way to move beyond identifying sensitive content in Confluence. The company said the product maps how Confluence data flows into AI systems. It also evaluates existing controls and highlights what information models could expose in response to a query.

Many enterprises already use SaaS applications for knowledge management and collaboration. Bedrock Data said the shift creates a security challenge because sensitive information sits outside traditional databases. The company also pointed to wider adoption of AI, which increases the number of systems capable of retrieving and summarising unstructured content.

Bedrock Data said it uses a read-only Atlassian access token for the integration. It said the approach keeps zero write permissions. It also said the model reduces operational risk during discovery and analysis.

"Confluence is one of the most requested SaaS data sources because, for many organisations, it holds vast amounts of internal knowledge, including trade secrets, intellectual property and customer information," said Bruno Kurtic, Co-Founder, President, and CEO, Bedrock Data. "This content is increasingly used across analytics, collaboration and AI use cases including as a basis for RAGs. Organisations need to ensure it is properly classified, protected and compliant with internal governance standards and regulatory requirements, with clear visibility into how sensitive information is accessed and used."

Mapping lineage

Bedrock Data said the product links SaaS discovery, sensitive data classification and AI inference lineage in a single platform. The company described its differentiator as mapping sensitive data back to the AI models that use it for inference. It said that the view shows what an AI model could surface in response to a query and whether an unauthorised user could access it.

Confluence content often sits behind nested permissions and inherited access paths. Bedrock Data said the integration analyses permissions at multiple levels and resolves indirect access. It said the process provides access that administrators may not see with space-level controls alone.

What it covers

Bedrock Data said the integration discovers Confluence spaces and maps content across folders, pages, live pages and blogs. It also said the platform scans unstructured text for PII, secrets and intellectual property. The company said it indexes metadata in a Bedrock Metadata Lake.

The company said security teams can query Confluence risk alongside other SaaS, cloud and AI data sources from the same interface. It also stated that the integration fits within a least-privilege security model through fine-grained, read-only access.

The announcement comes as security leaders focus on how enterprise knowledge bases integrate with AI tools. Many organisations have rolled out internal assistants that draw from Confluence pages and other repositories. That pattern increases the risk of an AI system returning sensitive content to a user with insufficient permissions, particularly when access controls do not translate cleanly into model responses.

Bedrock Data said its approach aims to show what information resides in Confluence and how it flows into AI inference pipelines. The company framed the integration as part of a broader expansion of DSPM into SaaS tools and AI workflows.

Roll-out

Bedrock Data said the Confluence integration is available to its customers immediately.

"Confluence is one of the most requested SaaS data sources because, for many organizations, it holds vast amounts of internal knowledge, including trade secrets, intellectual property and customer information," said Bruno Kurtic, Co-Founder, President, and CEO, Bedrock Data.