AI Security:

The Next Critical Layer of SASE

Rethinking Remote Access:  From Infrastructure to Interaction  Rethinking Remote Access: From Infrastructure to Interaction

AI Security: The Next Critical Layer of SASE | 2

Enterprise AI Transformation Artificial Intelligence is rapidly becoming embedded across modern enterprise environments.

Employees now interact daily with AI-powered SaaS applications, browser copilots, developer

assistants, autonomous agents, embedded generative AI services, and AI-enhanced business

workflows. AI capabilities are increasingly integrated into collaboration platforms, productivity suites,

customer engagement systems, browsers, and operational applications — often becoming part of

normal business processes without introducing separate user experiences.

AI now lives inside everyday work Employees touch AI continuously — usually without leaving the tools they already use.

SaaS copilots

Assistants built into

productivity & CRM suites

Browser copilots

AI search & extensions

running in the browser

Developer assistants

Code generation inside

IDEs and workflows

Autonomous agents Embedded generative AI AI-enhanced workflows

Acting on data and

triggering actions via

APIs

GenAI baked into the apps

people use all day

Automation that moves

data across business

processes

This rapid adoption is fundamentally reshaping how organizations operate, accelerating productivity

and automation across every business function. At the same time, it is introducing a new and highly

dynamic security challenge. Sensitive enterprise information now flows through prompts, uploads,

browser sessions, APIs, automated workflows, and AI-driven interactions that frequently operate

outside the visibility boundaries of traditional security architectures.

Many organizations already face growing exposure associated with unsanctioned AI usage, shadow AI

applications, AI-powered browser extensions, autonomous workflow automation, and external

generative AI platforms capable of processing sensitive corporate data. As AI adoption accelerates,

security teams must extend governance and protection consistently across an increasingly

distributed and rapidly evolving AI ecosystem.

This shift requires more than isolated AI security controls. Organizations need a unified architecture

capable of securing AI interactions wherever they occur — across users, devices, browsers, SaaS

applications, cloud environments, and enterprise workflows.

Enterprise AI Transformation

AI Security: The Next Critical Layer of SASE | 2

For many employees around the world, work does not start at the office. About 48% of the glo

bal workforce now works remotely (full or part-time), compared to 20% in the Pre-COVID era (

Capital Counselor Work from home statistics 2026 Report). Employees are working from publ

ic or semi-public locations like cafés, libraries, lobbies and lounges on a regular basis, meaning they ar

e highly mobile and not tied to working from home. For many enterprises, the defining charact

eristic of modern work is no longer remote versus office-based, but mobi

This rapid adoption is fundamentally reshaping how organizations operate, accelerating productivity

and automation across every business function. At the same time, it is introducing a new and highly

dynamic security challenge. Sensitive enterprise information now flows through prompts, uploads,

browser sessions, APIs, automated workflows, and AI-driven interactions that frequently operate

outside the visibility boundaries of traditional security architectures.

Many organizations already face growing exposure associated with unsanctioned AI usage, shadow AI

applications, AI-powered browser extensions, autonomous workflow automation, and external

generative AI platforms capable of processing sensitive corporate data. As AI adoption accelerates,

security teams must extend governance and protection consistently across an increasingly

distributed and rapidly evolving AI ecosystem.

This shift requires more than isolated AI security controls. Organizations need a unified architecture

capable of securing AI interactions wherever they occur - across users, devices, browsers, SaaS

applications, cloud environments, and enterprise workflows.

ntly untrusted, variable in performance, and invisible to traditional perimeter-based

le. These environments are inhere

security models. When it comes to access and sec

urity, traditional security systems assume they are either at the of fice, at home, or on a trusted network. Today, that is no longer

the case.  Employees require the same level of access, speed

and agility when accessing corporate resources, no matter where th ey are. Private and public files, SaaS work tools and resource s like Microsoft 365, Google Workspace and Salesforce must all be accessible without sacrificing security. Work no longer has a physical perimeter. It is carried wherever employees sit.

AI Security: The Next Critical Layer of SASE | 3

AI Introduces a New Enterprise Risk Model Enterprise data increasingly moves through AI-enabled environments in ways that are difficult to

monitor using traditional security models. Employees may unknowingly expose source code,

customer records, financial information, strategic plans, internal communications, or regulated data

to external AI services during normal business activity. AI-powered browser extensions and

embedded assistants further expand the attack surface by introducing additional third-party access

points into enterprise workflows.

Sensitive data now crosses the visibility boundary

Inside the enterprise Beyond enterprise control

Source code

Customer records

prompts, uploads,

API calls

External generative AI tools

Shadow AI applications

Financial information AI browser extensions

Strategic plans Autonomous AI agents

Regulated & personal data Third-party access points

The rapid proliferation of AI applications and AI agents has created a new operational and security

challenge: Shadow AI. Organizations need visibility into how AI tools are being used in real time, what

data is being shared, and which AI services are being accessed across the enterprise.

The emergence of autonomous AI agents introduces an additional layer of operational and security

complexity. Modern AI systems are increasingly capable of interacting directly with enterprise

applications, retrieving data, executing workflows, generating actions, and orchestrating business

processes through APIs and automation platforms. These environments require organizations to

govern both user-driven and AI-driven activity using consistent security policies and centralized

visibility.

The convergence of AI-driven productivity, automation, and autonomous operational capabilities is

fundamentally reshaping enterprise security requirements.

AI Introduces a New Enterprise Risk Model

For many employees around the world, work does not start at the office. About 48% of the global w

orkforce now works remotely (full or part-time), compared to 20% in the Pre-COVID era (Ca

pital Counselor Work from home statistics 2026 Report). Employees are working from public or semi-pub

lic locations like cafés, libraries, lobbies and lounges on a regular basis, meaning they a

re highly mobile and not tied to working from home. For many enterprises, the defining characterist

ic of modern work is no longer re

mote versus office-based, but mobile. These enviro

longer the case.  Employees require the same level of access, speed and agility when accessing corpo

rate resources, no matter where they are. Private and public files, SaaS work tools and resources like M

icrosoft 365, Google Workspace and Salesforce must all be accessible without sacrific

ing security. Work no longer has a physical perimeter. It is carried wherever employees sit.

The emergence of autonomous AI agents introduces an additional layer of operational and security co

mplexity. Modern AI systems are increasingly capable of interacting directly with enterprise

applications, retrieving data, executing workflows, generating actions, and orchestrating business

processes through APIs and automation platforms. These environments require organizations to govern both user-driven and AI-driven activity using consistent security policies and centralized visibility.

The convergence of AI-driven productivity, automation, and autonomous operational capabilities is

fundamentally reshaping enterprise security requirements.

AI Security: The Next Critical Layer of SASE | 3

nments are inherently untrusted, variable in p

erformance, and invisible to traditional perimet

security models. When it comes to access and se

curity, traditional security systems assum

e they are either at the office, at

home, or on a trusted network. Today, that is no

er-based

AI Security: The Next Critical Layer of SASE | 4

Security Architecture Requirements for the AI Era Traditional security architecture was not designed to govern highly distributed AI environments that

span browsers, SaaS applications, cloud services, APIs, remote users, embedded assistants, and

autonomous systems simultaneously. Many organizations currently rely on fragmented security tools

that provide only partial visibility into AI-related activity, creating inconsistent policy enforcement and

operational complexity.

From fragmented tools to one unified architecture

Today: fragmented point tools Unified SASE: one control plane

Check Point SASE SWG CASB

SWG CASB

DLP ZTNA

ZTNA DLP

Browser tool

Browser security Threat prevention

Gaps & blind spots

for AI activity Users · Devices · Browsers · SaaS ·

Cloud · APIs · AI workflows

One policy across

Securing AI-enabled enterprise environments requires an architecture capable of extending visibility,

governance, threat prevention, identity-aware access control, and data protection consistently across

the entire digital experience.

These requirements align naturally with the architectural foundations of Secure Access Service Edge

(SASE). SASE already operates at the convergence point of users, devices, browsers, SaaS

applications, cloud services, enterprise data, identity, and internet connectivity. This position enables

SASE to function as a centralized enforcement architecture for AI-driven environments without

requiring organizations to deploy disconnected point products across multiple operational domains.

A unified SASE architecture enables organizations to apply consistent governance policies across AI

applications, browser-based interactions, embedded AI services, autonomous workflows, and cloud-

delivered AI platforms while maintaining centralized operational control.

As AI adoption accelerates, extending AI security directly into the SASE architecture becomes

increasingly important for maintaining visibility, governance, compliance, and operational simplicity

across distributed enterprise environments.

Security Architecture Requirements for the AI Era

For many employees around the world, work does not start at the office. About 48% of the global workf

orce now works remotely (full or part-time), compared to 20% in the Pre-COVID era (Capital Coun

selor Work from home statistics 2026 Report). Employees are working from public or semi-public loc

ations like cafés, libraries, lobbies and lounges on a regular basis, meaning they are highly mobile and not

tied to working from h

istic of modern work is no longer remote versus office-based

ese env

ts are

tly untruste

nvisible to traditional perimeter-ba

sed security models. When it comes to access and se

, but mobile. Th

ironmen

inheren

d, variable in performance, and i

curity, traditional securit

y systems assume they are either at the office, at home, or on a trusted network. Today, that is no lo

nger the case.  Employees require the same level of access, speed and agility when accessing corporate resources, no matter where they are. Private and public files, SaaS work tools and resources like Microsoft 365, Google Workspace and Salesforce must all be accessible without sacrificing security. Work no longer has a physical perimeter. It is carried wherever employees sit.

Securing AI-enabled enterprise environments requires an architecture capable of extending visibility, governance, threat prevention, identity-aware access control, and data protection consistently across the entire digital experience.

These requirements align naturally with the architectural foundations of Secure Access Service Edge

(SASE). SASE already operates at the convergence point of users, devices, browsers, SaaS

applications, cloud services, enterprise data, identity, and internet connectivity. This position enables

SASE to function as a centralized enforcement architecture for AI-driven environments without

requiring organizations to deploy disconnected point products across multiple operational domains.

A unified SASE architecture enables organizations to apply consistent governance policies across AI

applications, browser-based interactions, embedded AI services, autonomous workflows, and cloud-

delivered AI platforms while maintaining centralized operational control.

As AI adoption accelerates, extending AI security directly into the SASE architecture becomes

increasingly important for maintaining visibility, governance, compliance, and operational simplicity

across distributed enterprise environments.

AI Security: The Next Critical Layer of SASE | 4

ome. For many enterprises, the defining character

AI Security: The Next Critical Layer of SASE | 5

Extending SASE Security to AI Environments Check Point SASE extends security across AI-enabled environments through integrated capabilities

that include Secure Web Gateway (SWG), CASB, Zero Trust Network Access (ZTNA), Data Loss

Prevention (DLP), browser security, threat prevention, and AI-powered threat intelligence. Together,

these capabilities provide organizations with centralized visibility and governance across AI-related

activity throughout the enterprise.

AI application visibility plays a critical role in this model. Security teams require the ability to identify

which AI services employees access, understand usage patterns, monitor interactions with

enterprise data, and evaluate exposure associated with sanctioned and unsanctioned AI applications.

Centralized visibility enables organizations to apply governance policies consistently while reducing

operational blind spots associated with shadow AI usage.

Data protection also becomes increasingly important as employees interact with external AI

platforms during normal business operations. Sensitive information such as source code, financial

records, customer information, intellectual property, regulated data, and strategic business content

frequently moves through prompts, uploads, and AI-driven workflows. Extending DLP enforcement

into AI environments helps organizations reduce the risk of unauthorized data exposure while

supporting secure AI adoption across the business.

Browser-based AI interactions introduce additional governance requirements. Many AI services

operate directly within enterprise browsers through embedded assistants, AI-enhanced search

platforms, browser extensions, and generative AI interfaces. Security teams therefore require

consistent visibility and policy enforcement across browser-driven AI activity as part of the broader

enterprise security architecture.

Agents

Employees Tool/ Privilege Abuse

Unsafe Autonomy

Loop/ DoW Risks

Applications

Shadow AI Usage

Sensitive
 Posts

Data Leakage

Prompt Injections

Harmful Outputs

Data Exfiltration

Discover

Protect

Govern

AI Defense Plane

One platform. One lens. From employees to applications to agents.

Extending SASE Security to AI Environments

For many employees around the world, work does not start at the office. About 48% of the global w

orkforce now works remotely (full or part-time), compared to 20% in the Pre-COVID era (Ca

pital Counselor Work from home statistics 2026 Report). Employees are working from public or semi-pub

lic locations like cafés, libraries, lobbies and lounges on a regular basis, meaning they are highly m

obile and not tied to working from

home. For many enterprises, the defining characteristic of modern work is no longer remote versus office-based

, but mobile. These environments are inherently untrusted, variable in performance, and i

nvisible to traditional perimeter-based security models. When it comes to access and security, tradi

tional security systems assume they are either at the office, at home, or on a trusted network. Today,

that is no longer the case.  Employees require the same

level of access, speed and agility when accessing corporate resources, no matter where the

y are. Private and public files, SaaS work tools and resources like Microsoft 365, Google Workspac

e and Salesforce must all be accessible without sacrificing security. Work no longer has a physical p

erimeter. It is carried wherever employees sit.

AI Security: The Next Critical Layer of SASE | 5

Discover

Protect

Govern

AI Defense Plane

One platform. One lens. From employees to applications to agents.

Shadow AI Usage

Sensitive Posts

Data Leakage

Prompt Injections

Harmful Outputs

Data Exfiltration

AI Security: The Next Critical Layer of SASE | 6

Threat prevention capabilities also play a critical role in protecting AI-enabled environments. Threat

actors increasingly leverage AI to accelerate phishing attacks, automate malicious content

generation, improve credential theft techniques, and enhance social engineering campaigns. AI-

powered threat intelligence and advanced threat prevention technologies help organizations identify

and block evolving AI-assisted threats before they impact enterprise environments.

Extending Zero Trust enforcement across AI-enabled workflows further strengthens security posture

by enabling organizations to apply identity-aware access policies consistently across users, devices,

SaaS applications, cloud services, APIs, and autonomous AI systems.

Architectural Framework for AI Security Within SASE The following architectural model illustrates how AI security capabilities integrate naturally into the

SASE architecture.

Users & Devices

Check Point SASE

SWG CASB ZTNA

Browser Security

AI Governance & Visibility

DLP

Threat Prevention

AI

AI SaaS
 Apps

Autonomous

Workflows

AI
 Browsers

Copilots

AI Agents & Chatbots

API-driven
 AI

This architecture enables organizations to extend centralized governance, visibility, and protection

consistently across AI-driven interactions while maintaining operational simplicity through a unified

cloud-delivered SASE platform.

Architectural Framework for AI Security Within SASE For many employees around the world, work does not start at the office. About 48% of the global workforc

e now works remote

ly (full or par

t-time), compare

d to 20% in th

e Pre-COVID era (Capital Coun selor Work from home statistic

s

2026 Report). Employe es are working from pu

blic or semi-public locations  like cafés,

libraries, lobbies and lounges on a regular basis, meaning they are highly mobile and not tied to wo

rking from home. For many enterprises, the defining characteristic of modern work is no longer remote versus office-based, but mobile. These environments are inherently untrusted, variable in performance, and invisible to traditional perimeter-based security models. When it comes to access and security, traditional security systems assume they are either at the office, at home, or on a trusted network. Today, that is no longer the case.  Employees require the same level of access, speed and agility when accessing corporate resources, no matter where they are. Private and public files, SaaS work tools and resources like Microsoft 365, Google Workspace and Salesforce must all be accessible without sacrificing security. Work no longer has a physical perimeter. It is carried wherever employees sit.

This architecture enables organizations to extend centralized governance, visibility, and protection consistently across AI-driven interactions while maintaining operational simplicity through a unified cloud-delivered SASE platform.

AI Security: The Next Critical Layer of SASE | 6

Threat prevention capabilities also play a critical role in protecting AI-enabled environments. Threat

actors increasingly leverage AI to accelerate phishing attacks, automate malicious content

generation, improve credential theft techniques, and enhance social engineering campaigns. AI-

powered threat intelligence and advanced threat prevention technologies help organizations identify

and block evolving AI-assisted threats before they impact enterprise environments.

Extending Zero Trust enforcement across AI-enabled workflows further strengthens security posture

by enabling organizations to apply identity-aware access policies consistently across users, devices,

SaaS applications, cloud services, APIs, and autonomous AI systems.

AI Security: The Next Critical Layer of SASE | 7

Securing the Future of Enterprise AI

AI capabilities continue to expand across enterprise infrastructure, business applications,

operational workflows, and cloud-delivered services. Organizations therefore require security

architectures capable of supporting large-scale AI adoption while maintaining visibility, governance,

compliance, data protection, and operational control.

SASE provides the architectural foundation required to secure modern AI-enabled enterprise

environments through the convergence of networking, security, identity, browser protection, SaaS

governance, Zero Trust enforcement, data protection, and AI-powered threat prevention within a

unified cloud-delivered architecture.

By extending AI security directly into the SASE framework, organizations can apply consistent

security controls across users, applications, browsers, SaaS platforms, cloud environments, APIs,

and AI-driven workflows without increasing operational fragmentation.

As enterprise AI adoption accelerates, integrating AI security into the SASE architecture will become

an increasingly important component of modern cybersecurity strategy.

Employee AI Interactions Covered by Check Point Workforce AI Check Point Workforce AI extends consistent governance and protection across the full range of AI

tools employees use every day — from public chat assistants to code assistants, desktop agents, and

autonomous workflows.

Employee AI interactions Covered by Check Point Workforce AI

Check Point Workforce AI Security Essentials Edition

Public AI tools e.g. ChatGPT, Gemini, Claude, Copilot

Browser extensions

Desktop agents

SaaS-integrated AI features

Code assistants e.g. GitHub Copilot, Cursor and MCP-connected workflows.

Check Point Workforce AI Security Enterprise Edition

All the features included in the Essentials Edition plus:

AI Security for Desktop AI Applications

AI Security for Code Assistance (IDEs)

AI Security for Agents

Inventory Endpoint Scanner for Agents including risk assessment

www.checkpoint.com  © 2026 Check Point Software Technologies Ltd. All rights reserved.

Employee AI Interactions Covered by Check Point Workforce AI

AI Security: The Next Critical Layer of SASE | 7

For many employees around the world, work does not start at the office. About 48% of the global wo

rkforce now works remotely (full or part-time), compared to 20% in the Pre-COVID era (Capital Counse

lor Work from home st

semi-public locations like cafés, libraries, lobbies and lounges on a regular basis, meaning they are high

ly mobile and not tied to working from home. For many enterprises, the defining ch

aracteristic of modern work is

no longer remote versus office-based, but mobile. These en

vironments are inherently untrusted, variable in perfo

rmance, and invisible to traditional perimeter-base

d security models. When it comes to access and security, traditional securi

ty systems assume they are either at the office, at home, or

atistics 2026 Report). Employees are working from public or

SASE provides the architectural foundation required to secure modern AI-enabled enterprise

environments through the convergence of networking, security, identity, browser protection, SaaS

governance, Zero Trust enforcement, data protection, and AI-powered threat prevention within a

unified cloud-delivered architecture.

By extending AI security directly into the SASE framework, organizations can apply consistent

security controls across users, applications, browsers, SaaS platforms, cloud environments, APIs,

and AI-driven workflows without increasing operational fragmentation.

As enterprise AI adoption accelerates, integrating AI security into the SASE architecture will become

an increasingly important component of modern cybersecurity strategy.

Securing the Future of Enterprise AI

AI capabilities continue to expand across enterprise infrastructure, business applications,

operational workflows, and cloud-delivered services. Organizations therefore require security

architectures capable of supporting large-scale AI adoption while maintaining visibility, governance,

compliance, data protection, and operational control.

on a trusted network. Today, that is no longer the case.  Employees require the same level of access, speed and agility when accessing corporate resources, no matter where they are. Private and public files, SaaS work tools and resources like Microsoft 365, Google Workspace and Salesforce must all be accessible without sacrificing security. Work no longer has a physical perimeter. It is carried wherever employees sit.