Prompt Flow Studio Adds Enterprise AI Data Controls

Prompt Flow Studio Adds Enterprise AI Security Controls Prompt Flow Studio Adds Enterprise AI Security Controls

EarthLink Network Co., Ltd. has launched an Enterprise plan for Prompt Flow Studio, adding administrator-controlled security policies designed to mask, flag or block sensitive information before it reaches external AI models. The platform introduces organization-wide data detection rules, audit trails, model restrictions, insider-risk signals and governance controls spanning AI providers including OpenAI, Anthropic and Google, positioning the workspace around enterprise AI security and compliance management.

Prompt Flow Studio Adds Enterprise Controls for AI Data Security

The Enterprise plan addresses a growing challenge for businesses adopting generative AI: how to allow employees to use multiple AI models while reducing the risk of confidential information, personal data or regulated content being inadvertently shared with external providers.

Prompt Flow Studio’s new controls operate before messages are transmitted to AI providers. Administrators can define organization-wide policies that members cannot disable, with enforcement options including masking, warnings and blocking. The company says these controls apply across supported AI models, rather than requiring separate security policies for individual providers.

Custom rules and AI-powered detection

The platform offers keyword-based detection, optional fuzzy matching and regular-expression rules for identifying information such as employee identifiers, internal case numbers and confidential terminology.

Administrators can assign different enforcement actions depending on the detected information. Rules can block messages outright, mask sensitive information or generate warnings, allowing organizations to tailor enforcement to the type of data and operational context.

Prompt Flow Studio also introduces AI-based detection for information that may not be captured by conventional keyword rules. Administrators can describe detection criteria in natural language, provide matching and non-matching examples, and test rules before deployment.

The testing process includes reported miss and false-alarm rates, helping administrators evaluate detection behavior before applying policies across their organization. The company acknowledges that AI-based detection is probabilistic, meaning the system cannot guarantee that every relevant message will be identified.

The approach combines conventional rule-based controls with AI-assisted detection, offering enterprises multiple mechanisms for identifying potentially sensitive content in employee prompts.

PII masking before AI model transmission

Prompt Flow Studio’s pre-send masking system is designed to identify and replace personal information before prompts reach AI providers.

The platform says it can mask email addresses, phone numbers, national identification numbers, payment card details, bank account numbers and API keys. Masked information is subsequently restored in the AI response, allowing users to work with protected information without transmitting the original identifiers to the model.

The masking capabilities can also extend to conversation history, attachments, organizational knowledge and fetched web pages.

However, the company identifies limitations: names, addresses and text embedded inside images are not covered by this masking feature. These exclusions are important for organizations evaluating the platform for regulated or sensitive workflows, where data protection requirements may extend beyond structured identifiers.

Audit trails and insider-risk monitoring

The Enterprise plan introduces event logging that records the user, timestamp, detection rule, AI provider and message location associated with each detection.

For certain blocked incidents, such as a user attempting to submit a message containing a restricted internal code or employee identifier, the system can record the event without storing the message itself.

The platform also supports administrator-defined warning rules for potential insider-risk scenarios. For example, an organization could create a rule identifying messages discussing resignation or job changes. Matching prompts can continue to the AI provider while the detection is logged for review.

The company says the platform does not independently identify departing employees or automatically classify these topics outside administrator-defined rules. Administrators also cannot open members’ conversations, according to the announcement.

The dashboard surfaces operational signals, including repeated PII detections, denied prompts, late-night activity and rule hits. These indicators are calculated from event records rather than message bodies, with optional anomaly detection capable of locking accounts when usage patterns appear unusual.

Governance, model restrictions and compliance workflows

Prompt Flow Studio’s enterprise controls extend beyond prompt protection to model access governance.

Administrators can restrict AI models based on developer country, allow providers from selected jurisdictions, block China-based developers collectively or restrict individual models. Blocked models are removed from the model selection interface and denied at runtime.

The platform also introduces Legal Hold functionality, which suspends deletion when records need to be preserved. Legal disclosures require re-authentication, approval and an audit trail.

EarthLink Network says sign-ins, permission changes, deletions, Legal Hold events and billing records are maintained for seven years using hash-chained, tamper-evident records.

The company also describes GDPR Article 17 erasure workflows, allowing administrators to erase personal data following re-authentication while retaining certain pseudonymized accounting and audit records under applicable legal retention provisions.

These features are intended to support organizations managing sensitive information and compliance requirements across enterprise AI workflows. Their suitability for specific regulatory obligations will depend on implementation, organizational policies and independent compliance assessment.

Enterprise AI governance moves beyond individual model security

Prompt Flow Studio’s Enterprise launch reflects a broader shift toward AI governance platforms that manage usage across multiple providers. As businesses introduce AI assistants into operations, employees may interact with several models through different interfaces, creating fragmented oversight and data-handling risks.

A centralized workspace with pre-send controls, auditability, usage monitoring and model restrictions could help organizations consolidate some of these governance requirements.

The platform separates seat licensing from AI usage, with administrators allocating shared AI tokens, called PFC, among members. This separates user access management from AI consumption allocation.

The commercial and technical proposition is therefore centered on controlling how employees interact with AI systems, rather than developing a standalone AI model.

Prompt Flow Studio’s effectiveness will depend on detection accuracy, coverage of supported data sources, administrator configuration, integration with enterprise workflows and how organizations balance security enforcement with employee productivity.

Market Landscape

Enterprise AI governance is evolving from individual model security toward centralized policy enforcement across multiple AI providers. Businesses increasingly need controls that manage data exposure, access restrictions, auditability and AI usage without completely restricting employee adoption.

Prompt Flow Studio’s Enterprise offering reflects several market trends:

  • AI data loss prevention (DLP): Pre-send masking and blocking help organizations limit sensitive information exposure through AI prompts.
  • Multi-model AI governance: Centralized policies across providers reduce the need to manage individual AI security controls separately.
  • AI compliance and auditability: Tamper-evident logs, Legal Hold and disclosure workflows address enterprise recordkeeping requirements.
  • Insider-risk monitoring: Policy-driven warnings and activity-based signals offer another layer of organizational risk monitoring.
  • AI access governance: Country-based restrictions and model-level controls give enterprises more control over which AI services employees can use.

The market includes established cybersecurity and data protection vendors such as Microsoft Purview, Netskope, Palo Alto Networks and other enterprise security providers expanding into AI governance and data protection.

Prompt Flow Studio is positioning itself as a multi-AI workspace with governance built into the interaction layer, rather than relying exclusively on separate enterprise security products. Its differentiation will depend on the reliability of its detection mechanisms, the scope of its protections, operational usability and how effectively it fits into existing enterprise security architecture.

Top Insights

  • Prompt Flow Studio has launched an Enterprise plan with organization-wide controls to mask, warn on or block sensitive information before AI transmission.
  • Administrators can combine keyword rules, regular expressions and AI-based contextual detection with testing for false alarms and missed detections.
  • The platform offers PII masking, audit logs, insider-risk signals, model restrictions and Legal Hold workflows.
  • Model access can be restricted by developer country or individual AI models, providing centralized multi-provider governance.
  • The Enterprise offering targets businesses seeking greater oversight of employee AI usage, confidential information and compliance-related workflows.

Power Tomorrow’s Intelligence — Build It with TechEdgeAI

Grow Your
Brand Visibility

Looking to publish a press release, guest article, interview or podcast? Connect with us.

GET FEATURED
Subscribe

Sign up today for exclusive insights and updates.

Newsletter Signup