AI discovery and technical security assessment
AI discovery and data-exposure assessment to prioritize technical safeguards
Understand the AI use and integrations within your agreed scope, including where sensitive data can flow. ICA’s discovery and configuration review connects an AI inventory with exposure findings and prioritized technical recommendations.
MIS
CISSP
CHISL
CPHIMS
CDH-E
Will Long Sr. · Founder
MIS · CHISL · CDH-E · CISSP · CPHIMS
Nearly 40 years in IT & cybersecurity
Former VP & CISO, Children’s Health System of Texas
Tim Pelikan · Co-Founder
Credential details pending confirmation
30 years in cybersecurity & data risk
Interviews & industry conversations
Coffee With Closers
Cybersecurity With Discern
“ICA provides distinctive security professionals who excel at building strong client relationships, demonstrate exceptional organization, and deliver practical, risk-based consulting recommendations to our clients.”
Michael Cox, CIPP/US
Chief Privacy Consultant & Founder, SoCal Privacy Consultants
Partner endorsement of ICA’s advisory work
Make AI rollout decisions with a clearer exposure picture
A new copilot or agent is being introduced, but its data access is unclear. Teams may use unrecorded AI tools. Security and privacy leaders need evidence to prioritize technical safeguards.
This assessment is intended for CISOs, CIOs, AI platform owners and privacy leaders facing those decisions. It focuses on identifying agreed AI use and examining its technical exposure. Findings give platform owners a basis for discussing access changes and data-handling priorities with leadership.
Define the AI tools and data connections to review
Coverage starts with the systems and AI integrations your organization agrees to include. The review considers how users and connected services gain access, which permissions apply and where data moves. Relevant context may include copilots, agents or AI features within existing applications.
For each included workflow, establish the business purpose and the data it handles. Review available evidence of connections to other systems and the access granted through those connections. Discovery depends on the agreed environment and accessible evidence; it should not be treated as proof that every AI tool across the organization has been found.
Trace access and data flows through available evidence
The engagement uses scoped discovery and configuration review. Discussions with platform and integration owners help explain how the included AI workflows operate. Available inventories, configuration records and permission information provide the evidence for reviewing technical exposure.
The review distinguishes documented settings from areas that cannot be verified. An integration described in a diagram, for example, still needs supporting access information before its permissions can be assessed. Tool access and evidence depth are confirmed during scoping. This method does not imply vulnerability scanning or authorized penetration testing; any active testing requires its own validated scope and written authorization.
Use the findings to direct technical work
The catalog identifies three typical outputs, with their detail agreed in the engagement scope:
AI inventory. A record of the AI use and integrations identified within the agreed boundary. Platform owners can use it to reconcile known tools with the evidence reviewed and identify visibility gaps.
Exposure findings. Observations about access, permissions and data flows, supported by available evidence. These help security and privacy stakeholders understand where sensitive-data handling needs closer attention.
Prioritized technical recommendations. Suggested safeguards and technical changes in priority order. Delivery owners can use these to plan work and discuss dependencies before changing production settings.
Recommendations guide decisions. They do not establish that changes have been implemented or that an exposure has been resolved.
Prepare the environment context and evidence access
Scoping establishes the assessment objective, included systems and integrations, relevant data, and available tool access. The number of environments and the depth of configuration review influence effort and timing. Evidence availability also matters, particularly when different teams own the AI service and its connected data sources.
Identify people who can explain the included workflows and authorize access to relevant records. Useful preparation may include an existing AI inventory, integration diagrams and permission exports where available. Agree appropriate evidence handling and access arrangements before work begins. Supported tools and specialist fit must be confirmed for the proposed environment rather than assumed from a platform name.
Keep discovery separate from governance and adversarial testing
This assessment examines technical visibility and exposure. The AI & Agentic AI Governance Assessment addresses the separate question of organizational accountability and oversight. Technical findings can inform that work, but this engagement does not substitute for the MAT-10 governance review.
Model red teaming and prompt-injection testing are separate unless validated and expressly scoped. Managed monitoring has the same separate-scope requirement. The proposed assessment is a defined review, not an ongoing detection service. Remediation execution also requires a separate agreement. Findings reflect the systems and evidence reviewed and cannot guarantee that AI use is secure or compliant.
AI discovery and exposure assessment FAQs
Who should consider this assessment?
It fits security, IT, AI platform and privacy leaders who need clearer visibility into AI use or sensitive-data exposure. A new copilot or agent, an incomplete inventory, or uncertainty about integration permissions can justify a focused review.
Will it identify every unapproved AI tool?
Coverage depends on the systems and evidence agreed for discovery. The inventory records what is identified within that boundary. Evidence gaps and inaccessible environments should remain visible so leadership can decide whether further discovery is needed.
Does the review include attacking our AI systems?
The proposed method is discovery and configuration review. Model red teaming and prompt-injection testing require separate validation and scope. Any intrusive testing also needs written authorization and agreed rules of engagement for the authorized assets.
What determines cost and timing?
Included systems, integration complexity and review depth affect the engagement. Available tool access and the time needed to obtain evidence also matter. Those factors need agreement before price or timing can be confirmed.
How does this differ from an AI governance assessment?
TEC-20 focuses on identified AI use, access and technical data exposure. MAT-10 reviews governance and oversight. An organization may need both perspectives, but each engagement retains its own scope and outputs.
Are remediation and continuing monitoring included?
The typical outputs are an AI inventory, exposure findings and prioritized technical recommendations. Implementation requires separate agreement. Managed monitoring is separate unless validated and scoped; the assessment does not provide continuous visibility after the review.
Set a focused scope for AI exposure decisions
Start with the AI rollout or data-access concern your team needs to resolve. Define the relevant environments and evidence so the inventory and recommendations address that decision. Discuss a scoped review to prioritize technical safeguards.
Make AI rollout decisions with a clearer exposure picture
A new copilot or agent is being introduced, but its data access is unclear. Teams may use unrecorded AI tools. Security and privacy leaders need evidence to prioritize technical safeguards.
This assessment is intended for CISOs, CIOs, AI platform owners and privacy leaders facing those decisions. It focuses on identifying agreed AI use and examining its technical exposure. Findings give platform owners a basis for discussing access changes and data-handling priorities with leadership.
Define the AI tools and data connections to review
Coverage starts with the systems and AI integrations your organization agrees to include. The review considers how users and connected services gain access, which permissions apply and where data moves. Relevant context may include copilots, agents or AI features within existing applications.
For each included workflow, establish the business purpose and the data it handles. Review available evidence of connections to other systems and the access granted through those connections. Discovery depends on the agreed environment and accessible evidence; it should not be treated as proof that every AI tool across the organization has been found.
Trace access and data flows through available evidence
The engagement uses scoped discovery and configuration review. Discussions with platform and integration owners help explain how the included AI workflows operate. Available inventories, configuration records and permission information provide the evidence for reviewing technical exposure.
The review distinguishes documented settings from areas that cannot be verified. An integration described in a diagram, for example, still needs supporting access information before its permissions can be assessed. Tool access and evidence depth are confirmed during scoping. This method does not imply vulnerability scanning or authorized penetration testing; any active testing requires its own validated scope and written authorization.
Use the findings to direct technical work
The catalog identifies three typical outputs, with their detail agreed in the engagement scope:
AI inventory. A record of the AI use and integrations identified within the agreed boundary. Platform owners can use it to reconcile known tools with the evidence reviewed and identify visibility gaps.
Exposure findings. Observations about access, permissions and data flows, supported by available evidence. These help security and privacy stakeholders understand where sensitive-data handling needs closer attention.
Prioritized technical recommendations. Suggested safeguards and technical changes in priority order. Delivery owners can use these to plan work and discuss dependencies before changing production settings.
Recommendations guide decisions. They do not establish that changes have been implemented or that an exposure has been resolved.
Prepare the environment context and evidence access
Scoping establishes the assessment objective, included systems and integrations, relevant data, and available tool access. The number of environments and the depth of configuration review influence effort and timing. Evidence availability also matters, particularly when different teams own the AI service and its connected data sources.
Identify people who can explain the included workflows and authorize access to relevant records. Useful preparation may include an existing AI inventory, integration diagrams and permission exports where available. Agree appropriate evidence handling and access arrangements before work begins. Supported tools and specialist fit must be confirmed for the proposed environment rather than assumed from a platform name.
Keep discovery separate from governance and adversarial testing
This assessment examines technical visibility and exposure. The AI & Agentic AI Governance Assessment addresses the separate question of organizational accountability and oversight. Technical findings can inform that work, but this engagement does not substitute for the MAT-10 governance review.
Model red teaming and prompt-injection testing are separate unless validated and expressly scoped. Managed monitoring has the same separate-scope requirement. The proposed assessment is a defined review, not an ongoing detection service. Remediation execution also requires a separate agreement. Findings reflect the systems and evidence reviewed and cannot guarantee that AI use is secure or compliant.
AI discovery and exposure assessment FAQs
Who should consider this assessment?
It fits security, IT, AI platform and privacy leaders who need clearer visibility into AI use or sensitive-data exposure. A new copilot or agent, an incomplete inventory, or uncertainty about integration permissions can justify a focused review.
Will it identify every unapproved AI tool?
Coverage depends on the systems and evidence agreed for discovery. The inventory records what is identified within that boundary. Evidence gaps and inaccessible environments should remain visible so leadership can decide whether further discovery is needed.
Does the review include attacking our AI systems?
The proposed method is discovery and configuration review. Model red teaming and prompt-injection testing require separate validation and scope. Any intrusive testing also needs written authorization and agreed rules of engagement for the authorized assets.
What determines cost and timing?
Included systems, integration complexity and review depth affect the engagement. Available tool access and the time needed to obtain evidence also matter. Those factors need agreement before price or timing can be confirmed.
How does this differ from an AI governance assessment?
TEC-20 focuses on identified AI use, access and technical data exposure. MAT-10 reviews governance and oversight. An organization may need both perspectives, but each engagement retains its own scope and outputs.
Are remediation and continuing monitoring included?
The typical outputs are an AI inventory, exposure findings and prioritized technical recommendations. Implementation requires separate agreement. Managed monitoring is separate unless validated and scoped; the assessment does not provide continuous visibility after the review.
Set a focused scope for AI exposure decisions
Start with the AI rollout or data-access concern your team needs to resolve. Define the relevant environments and evidence so the inventory and recommendations address that decision. Discuss a scoped review to prioritize technical safeguards.