live wire
▸MotherDuck adds image classification and object location to prompt_jev(), returning typed SQL outputs with confidence scoresMotherDuck▸BigQuery conversational analytics adds Montréal and Toronto; mixed-region agents still default to globalGoogle Cloud Documentation▸Fabric enables Azure OpenAI Copilot access by default, but OpenAI-subprocessor and cross-region processing/storage switches remain offMicrosoft Learn▸Snowflake Cortex Agents now continue by default when named tools are inaccessible, emitting warnings instead of failing the runSnowflake documentation▸Databricks Genie One can use group memberships to tailor answers, while Unity Catalog remains the access-control boundaryDatabricks documentation▸Fabric data agents can query Power BI models with Read alone; Build and workspace access are unnecessaryMicrosoft Learn▸Fabric data-agent Git configs expose column flags, but selecting a table still includes every columnMicrosoft Learn▸Google Data Agent Kit offers remote Spark kernels, but notebook exports can still move results into local Pandas or ArrowGoogle Cloud Documentation▸Cube says production analytics agents need explicit partial-result states, caller-filtered model search and durable artifactsCube▸Thinking Machines names Bridgewater behind ReViSQL-K2.6, citing 92.97% SC-16 accuracy at 15% of frontier-model costThinking Machines Lab▸AWS says a mismatched data/tool/prompt release cut its Field Sales agent pass rate from 78% to 38% in pre-productionAWS Artificial Intelligence Blog▸NL2SQL Multi-Agent ships evaluation artifacts across BIRD, CSpider, Archer and KDD DataSpace—but no software licenseGitHub▸Snowflake says Cortex Analyst apps keep working, but recommends Cortex Agents; semantic views and verified queries carry overSnowflake Documentation▸IBM watsonx.governance adds agent runtime-control sync plus Langfuse and Arize metric ingestionIBM watsonx documentation▸MotherDuck adds image classification and object location to prompt_jev(), returning typed SQL outputs with confidence scoresMotherDuck▸BigQuery conversational analytics adds Montréal and Toronto; mixed-region agents still default to globalGoogle Cloud Documentation▸Fabric enables Azure OpenAI Copilot access by default, but OpenAI-subprocessor and cross-region processing/storage switches remain offMicrosoft Learn▸Snowflake Cortex Agents now continue by default when named tools are inaccessible, emitting warnings instead of failing the runSnowflake documentation▸Databricks Genie One can use group memberships to tailor answers, while Unity Catalog remains the access-control boundaryDatabricks documentation▸Fabric data agents can query Power BI models with Read alone; Build and workspace access are unnecessaryMicrosoft Learn▸Fabric data-agent Git configs expose column flags, but selecting a table still includes every columnMicrosoft Learn▸Google Data Agent Kit offers remote Spark kernels, but notebook exports can still move results into local Pandas or ArrowGoogle Cloud Documentation▸Cube says production analytics agents need explicit partial-result states, caller-filtered model search and durable artifactsCube▸Thinking Machines names Bridgewater behind ReViSQL-K2.6, citing 92.97% SC-16 accuracy at 15% of frontier-model costThinking Machines Lab▸AWS says a mismatched data/tool/prompt release cut its Field Sales agent pass rate from 78% to 38% in pre-productionAWS Artificial Intelligence Blog▸NL2SQL Multi-Agent ships evaluation artifacts across BIRD, CSpider, Archer and KDD DataSpace—but no software licenseGitHub▸Snowflake says Cortex Analyst apps keep working, but recommends Cortex Agents; semantic views and verified queries carry overSnowflake Documentation▸IBM watsonx.governance adds agent runtime-control sync plus Langfuse and Arize metric ingestionIBM watsonx documentation
nl2sql.ai
guideADMIN GUIDE

Fabric’s AI admin review needs five switches—not one Copilot toggle

Microsoft enables the main Azure OpenAI access setting by default, while provider access, two cross-region processing controls and storage remain separate. Approval should record five settings across four decisions.

By The News Desk· Oct 18, 2026the quick take — two AI hosts go live when you do

Microsoft Fabric does not have one master decision for AI access and data residency. Its current admin documentation describes a set of independent tenant controls, with different defaults and scopes. That matters for data agents because an approval that records only “Copilot enabled” leaves several material choices undocumented.

Start with the default posture

Microsoft says Users can use Copilot and other features powered by Azure OpenAI is enabled by default. The separate setting that permits OpenAI-operated models, with OpenAI acting as a Microsoft subprocessor, is disabled by default. Cross-region processing switches for Azure OpenAI and for OpenAI-operated models are also disabled by default, as is the switch that permits Azure OpenAI data to be stored outside the capacity’s region, compliance boundary or national-cloud instance.

Those are not interchangeable controls. The first governs whether users can reach Azure OpenAI-powered Fabric features. Another governs whether OpenAI-operated models are available. Two more govern where processing may occur, depending on the provider. A further switch governs cross-region storage for supported experiences, including Fabric data agents.

Treat processing and storage separately

A residency review should therefore record at least four decisions:

  1. Feature access: who may use Copilot and agent features powered by Azure OpenAI.
  2. Model provider: whether OpenAI-operated models are allowed as a Microsoft subprocessor.
  3. Processing location: whether either provider may process data beyond the capacity’s geography or compliance boundary.
  4. Storage location and retention: whether supported agent experiences may persist conversation history outside that boundary.

Microsoft says conversational experiences such as Fabric data agents can retain cross-session conversation history for up to 28 days unless the user clears it. It also distinguishes those experiences from ones that do not preserve history across sessions. So a security review should test both the tenant setting and the user-facing deletion path; enabling AI does not answer the retention question.

The entry point adds another boundary

Microsoft also warns that a Fabric data agent can be consumed from Foundry, Copilot Studio, Microsoft 365 Copilot or as an MCP server. Responses sent to those non-Fabric services may leave Fabric’s compliance boundary or geographic region and are then handled under the receiving service’s terms and data policies.

That means a compliant configuration inside Fabric is not the end of the review. Inventory every allowed consumer, map its processing and storage terms, and test the effective policy from that entry point.

The practical release artifact is a short control record: enabled user groups, permitted model providers, processing regions, storage and retention policy, and approved external consumers. Re-check it whenever an administrator enables a new provider, region exception or agent integration. Fabric exposes these as separate switches; the change process should preserve that separation.

sources

Filed by The News Desk. Corrections: desk@nl2sql.ai · Our standards →

comments · 0

    Comments are moderated before they appear. Your email is used once to confirm it is you — never shown, never sold. Corrections and questions get an answer from the desk when we have one.

    more from the desk