FAQs about AI Agents

What’s the difference between LLMs and AI Agents?

LLM (Large Language Model) is an AI model trained on a vast amount of data to recognize and generate natural language. It functions as a “language brain,” predicting each next word in a sentence. However, traditional LLMs are limited to their initial training data, lack the ability to interact with the outside world, and cannot update themselves with new information after training.

AI Agent, on the other hand, is a much more complete system that typically uses an LLM as its core intelligence foundation but is supplemented with sensors (gathering information), actuators (performing actions), knowledge bases (storing knowledge), and control systems (making decisions). This structure allows AI Agents to not only understand language but also interact with the surrounding environment.

The decisive difference between LLMs and Agents lies in the AI Agent’s “tool calling” capability. Through this mechanism, an Agent can retrieve updated information from external sources, optimize workflows, automatically break down and solve subtasks, store interactions in long-term memory, and plan for future actions. These capabilities help AI Agents provide more personalized experiences and comprehensive responses, while expanding their practical application across many fields.

>>> EXPLORE: Are AI Agents the next frontier of NLP Chatbots?

Are Reasoning models (like OpenAI o3 and DeepSeek R1) AI agents?

No. Reasoning models like o3 and R1 are LLMs trained to reason through solutions to complex problems. They do this by breaking them down into multiple steps using chain of thought. These LLMs cannot naturally interact with other systems or extend their reasoning beyond architectural limitations.

How do AI Agents integrate with existing systems and workflows?

The most common ways to integrate AI agents are:

• Connect it to a RAG platform, providing native connections between LLMs and knowledge bases. This allows the agent to use representations of your documents and business data as context for future responses, increasing output accuracy.
• Through APIs to external services. When you configure function calls in the AI agent platform, the model interacts with API endpoints in the same way a traditional program would, creating all the headers and body of the call.

How does Human-in-the-loop fit into the AI Agent workflow?

Human-in-the-loop frameworks enhance supervision of AI agent systems. Simply put, the agent’s actions pause at predetermined points in the workflow. A notification is sent to the user, who must review decisions, information, and scheduled tasks. Based on this information, the user will approve or change how the AI agent will continue the task.

Will AI agents take our jobs?

This technology will certainly replace jobs and bring changes to the market, although there is no clear vision of when and how this might happen. Workers may be replaced by AI agents in many industries. At the same time, many positions for AI development and maintenance may be created, along with human-in-the-loop positions, to ensure that human decisions control AI actions rather than the other way around.

Do AI agents exacerbate bias and discrimination?

An AI model is only as unbiased as the data it was trained on—so yes, they are biased. Addressing these issues involves changing machine learning processes and creating datasets that represent the full spectrum of the world and human experiences.

Who is responsible when an AI Agents makes a mistake?

A difficult problem in ethics and law, it’s still unclear who should be blamed for accidents and unintended consequences. The developers? Hardware/software owners? Operators? As new laws are created and industry barriers are implemented, we will be able to understand what roles AI agents can—and cannot—assume.

In short, with the ability to be autonomous, operate independently, make decisions based on data and real-world environments, AI Agents are a powerful automation solution that helps businesses optimize processes. The AI ​​Agents market is forecast to reach a size of 30 billion USD by 2033 and maintain a growth rate of about 31% per year.

The explosive potential of this technology in the next decade is huge. Contact FPT.AI now to take advantage of the enormous power of AI colleagues, accelerate innovation, enhance customer experience and scale more efficiently than ever!

Contact FPT.AI now for more information and a free consultation:

Hotline: 1900 638 399
Website: www.fpt.ai
Address: FPT Tower, 10 Pham Van Bach Street, Dich Vong Ward, Cau Giay District, Hanoi

>>> Read more about:

• 2025 technology trends: The explosive development of AI Agents
• Generative AI vs Machine Learning: Key Differences

2025 marked a pivotal milestone for FPT.AI, as this “Make in Vietnam” AI platform achieved strong growth in scale and technological capability, while also solidifying its presence on the regional and global AI map.

From expanding to more than 200 global customers in 15 countries and serving over 20 million end users, to the rapid proliferation of AI Agents, 2025 is the year AI truly transitions from a “visionary concept” into an essential technology embedded in enterprise operations. Let’s revisit some notable highlights of FPT.AI in 2025.

Proven technological excellence through key products, solutions, and strategic projects

Developing more than 70 language models deployed in real-world applications, addressing industry-specific challenges, and processing more than 1.111 trillion tokens.

FPT Smart Cloud GenAI Product Center has researched, trained, and integrated more than 70 large language models into real-world products. These models have processed 169 million requests, utilizing 1.111 trillion tokens to solve complex, industry-specific inquiries.

10,220 active AI agents in business operations

An ecosystem of 10,220 AI Agents has been deployed to help enterprises automate customer service, sales, internal operations, and data analytics. Each AI Agent functions as a dedicated “digital workforce,” operating tirelessly 24/7 across multiple channels.

FPT AI Mentor expands its footprint in the Japanese market

Mishima Kosan Co., Ltd., a leading Japanese manufacturing enterprise, has partnered with FPT to integrate AI into its corporate training. The FPT AI Mentor solution enables Mishima Kosan to standardize training programs, shorten skill-upgrading cycles, and support employees in a multilingual working environment. This strategic collaboration has significantly strengthened FPT.AI’s presence across more than 15 countries worldwide.

Currently, over 50,000 employees are trained each month through FPT AI Mentor.

FPT AI eKYC processed 75 million identity verifications, digitalizing customer onboarding for banks, fintech companies, and insurance firms with high accuracy, speed, and strict compliance with biometric anti-spoofing and security standards.

FPT AI Read processed more than 12 million documents.

By processing 12 million documents, FPT AI Read enables enterprises to eliminate manual data entry, rapidly reading and extracting information from contracts, records, invoices, and other business documents with high precision.

FPT AI Enhance quality-checked (QC) over 70 million customer service calls, enabling enterprises to monitor customer service quality at scale, detect errors, identify improvement opportunities, and ensure compliance with standardized scripts and tone of voice.



Launch of 03 new products and solutions

2025 was a particularly productive year for the FPT.AI product development team, marked by the introduction of three new solutions with outstanding capabilities:

• FPT AI Voice Agents empower enterprises to build AI-powered contact center teams. These Voice Agents are capable of handling up to 50 million calls per month, equivalent to a large-scale call center, operating 24/7 without human resource constraints. As a result, businesses can automate up to 70% of call center workloads, allowing human agents to focus on complex cases, improve customer experience, and reduce operating costs.
• FPT AI Adjust – AI–powered claims assistant, specializing in the insurance industry. The solution automates insurance claim intake and processing, enabling faster, more accurate case handling while reducing manual workload and improving customer satisfaction.
• AI Solutions for Smart City, supporting real-time analysis and monitoring of transportation infrastructure construction, traffic conditions, and incident alerts, while delivering in-depth reports and analytics for authorities.

Vietnamese AI workforce serving the nation’s transformation in the rising era

FPT AI Agents accompany the Nation into the AI Era In 2025, the “Make in Vietnam” AI workforce played a strategic role in supporting national digital transformation. FPT AI Agents accompanied officials from the General Statistics Office (Ministry of Finance), Ho Chi Minh City Tax Department, Ho Chi Minh City Department of Construction, and Ca Mau Provincial Employment Service Center.

Also, FPT.AI’s product ecosystem was selected by the Ministry of Science and Technology as a representative portfolio of innovative, digital transformation technologies and officially published on the Ministry’s electronic portal. The FPT.AI ecosystem includes FPT.AI, FPT AI Agents, FPT AI Mentor, FPT AI Knowledge Explore, and FPT AI Engage, was officially recognized by the Ministry of Science and Technology as exemplary science, technology, innovation, and digital transformation solutions

with strong application potential and strategic significance. The electronic portal curates and rigorously evaluates a ‘Make in Vietnam’ ecosystem, affirming Vietnam’s technological self-reliance and autonomy in a context where science and technology are defined as key drivers of national development. Official information is now available at nq57.mst.gov.vn.

Prestigious awards and global certifications



Scored a double win for best AI solutions at AI Awards 2025

Affirming its position as a leading technology company, FPT Smart Cloud achieved two major wins at AI Awards 2025, including Best AI Solution with FPT AI Agents Platform and Outstanding AI Enterprise.

FPT.AI honored at ASOCIO Awards

Surpassing numerous international competitors, FPT.AI became the first AI platform from Vietnam to be recognized by the Asian-Oceanian Computing Industry Organization (ASOCIO) in the “AI Service Provider” category, one of the most prestigious technology awards in the Asia-Pacific region.

FPT.AI’s talented AI engineers won 4 awards at VLSP 2025

Their outstanding performance at VLSP 2025 – the largest academic arena for Vietnamese language processing – once again reaffirms the research excellence of the FPT.AI team

FPT AI eKYC certified to ISO/IEC 30107-3 level 2

After passing over 400 sophisticated spoofing attack tests, including 3D masks, silicone, video replays, and projection attacks, FPT AI eKYC successfully met the ISO/IEC 30107-3 Level 2 international security standards. Certified by FIME, a globally renowned testing laboratory, the solution achieved an APCER of 0.75% (below the benchmark <2.5%) and a live face recognition rate exceeding 98%.

These milestones represent key highlights within the digital transformation landscape that FPT.AI continues to advance alongside enterprises and public sectors. The year 2025 proves that AI is no longer a distant future trend, but a core infrastructure for driving growth, operational optimization, and enhancing the customer experience. FPT.AI’s journey will continue with more remarkable breakthroughs ahead.

Continuing Generative AI, Agentic AI emerges as a promising technology, ushering in a new era of autonomy and independent thinking for artificial intelligence applications. What exactly is Agentic AI, and how does it differ from Generative AI? Let’s explore with FPT.AI in the following article.

What is Agentic AI?

As a new branch of artificial intelligence, Agentic AI represents systems equipped with Reinforcement Learning, language self-awareness, and contextual decision-making capabilities. It can analyze, set goals, strategize, make decisions, and even adjust its behavior through trial-and-error mechanisms to ensure goal achievement without continuous human intervention.

One of Agentic AI’s standout features is its ability to “chain tasks”—breaking down complex missions into smaller, manageable steps that can be efficiently executed. For instance, when tasked with creating a website, an AI agent system can autonomously plan and execute steps such as:

1. Designing the website structure and layout.
2. Drafting content for individual pages.
3. Writing necessary HTML, CSS, and backend code.
4. Designing graphics and integrating visuals.
5. Testing responsiveness and fixing any issues.



>>> Explore: What Are AI Agents? The Difference Between AI Agents and AI Chatbots

How Does Agentic AI Work?

Agentic AI operates through a four-step process to effectively and autonomously solve problems:

1. Perceive: It gathers and processes data from various sources like sensors, databases, and digital interfaces. This step involves extracting meaningful features, identifying objects, and recognizing relevant entities to build a comprehensive understanding of the surrounding context.
2. Reason: Large Language Models (LLMs) serve as reasoning engines, understanding tasks, generating solutions, and collaborating with specialized models to perform functions like content creation, image processing, or recommendation systems. Techniques such as Retrieval-Augmented Generation (RAG) are used to access proprietary data sources, enabling precise and contextually relevant results.
3. Act: By integrating with external tools and software through APIs, Agentic AI quickly executes tasks based on predefined plans. Control measures ensure accurate task completion, such as capping spending limits for customer service agents.
4. Learn: Through a continuous feedback loop called the Data Flywheel, Agentic AI learns from its interactions. Data is constantly collected and analyzed to refine models, adapt strategies, and enhance decision-making over time, thereby improving operational efficiency.



The Difference Between Agentic AI and Generative AI

Agentic AI surpasses both traditional AI systems and Generative AI (GenAI). Below is a comparison to clarify the differences:

Feature	Traditional AI	Generative AI (GenAI)	Agentic AI
Core Capability	Automates repetitive tasks with fixed rules. Relies on human programming and oversight.	Creates new content like text, images, audio, and code. Operates based on human commands, enhancing creativity and efficiency.	Acts autonomously, setting goals, strategizing, making decisions, and executing actions independently.
Autonomy	Lacks autonomy; follows pre-programmed rules.	Limited autonomy; depends on human guidance for context and output.	High autonomy; makes decisions and acts without continuous human intervention.
Learning	Cannot learn; executes pre-set rules.	Learns from large datasets to improve results but requires human prompts to act.	Learns from experience, adapts over time, and optimizes actions and goals autonomously.
Responsiveness	Predefined responses, static reactions.	More flexible responses based on context but relies on user commands and training data.	Proactively adapts strategies and goals based on situational context and objectives.
Decision-Making	Limited, rule-based.	Suggests ideas based on prompts.	Sets goals and makes contextual decisions independently.

While both technologies fall under artificial intelligence, they cater to distinct purposes. Agentic AI focuses on autonomous actions and complex task execution without continuous oversight, while GenAI relies on human input to determine the context and goals for creative outputs.

The core distinction lies in their outputs: GenAI produces creative content, while Agentic AI delivers actionable solutions and decisions. Together, they can form powerful systems that combine creativity with automation, enhancing efficiency across diverse applications.

Real-World Applications of Agentic AI

Agentic AI holds vast potential for applications across various industries. Here are some notable examples:

1. Business Operations: Handles massive datasets, automates supply chain management, optimizes inventory levels, forecasts demand, and manages complex logistics plans in real-time, enhancing efficiency and reducing costs.
2. Healthcare: Functions as 24/7 health assistants, personalizing treatment plans and offering proactive care by predicting potential health issues through Big Data analysis.
3. Software Development: Manages the entire software development lifecycle, from designing architecture and coding to debugging and quality assurance, revolutionizing how digital products are built and maintained.
4. Cybersecurity: Monitors network traffic, detects anomalies, and addresses real-time threats without human oversight, allowing experts to focus on complex challenges.
5. Human Resources: Automates tasks like candidate screening, interview scheduling, onboarding, and training while providing personalized career development advice.
6. Scientific Research: Accelerates breakthroughs by automating experiments, analyzing results, and formulating new hypotheses, driving innovation across various scientific fields.
7. Finance: Revolutionizes portfolio management, market trend analysis, and real-time investment strategies, enabling higher returns for investors.



Why Is Agentic AI a Promising Technology Trend?

Agentic AI is gaining significant attention in the tech world for the following reasons:

• High Autonomy: Operates independently, making it invaluable for tasks requiring constant monitoring or rapid decision-making.
• Problem-Solving Capabilities: Combines machine learning with goal-oriented behavior to provide innovative and efficient solutions.
• Flexibility and Adaptability: Adjusts strategies dynamically based on new information or environmental changes.
• Personalization: Delivers tailored solutions by learning from previous interactions, enhancing customer service processes.
• Scalability: Deployable across various applications and industries after training, enabling rapid and effective transformation.
• Enhanced Communication: Processes natural language effectively, facilitating seamless human interaction and control.



Challenges in Deploying Agentic AI

Despite its potential, implementing Agentic AI poses significant challenges:

1. Security Risks: Independent decision-making can expose systems to vulnerabilities, such as adversarial attacks, requiring robust security measures.
2. Unintended Behaviors: Autonomy may lead to unforeseen actions, necessitating transparent decision-making frameworks for effective management.
3. Resource Costs: Large-scale Agentic AI models demand substantial energy and data, raising sustainability concerns.
4. Ethical and Social Issues: Concerns include job automation, accountability for AI-driven decisions, and biases in training data affecting outcomes in sensitive areas like law enforcement or recruitment.
5. Human Oversight: Developing safety mechanisms like kill switches to control autonomous systems is complex, raising concerns about alignment with human values.

In conclusion, Agentic AI offers vast potential for enhancing efficiency and automation across industries. However, balancing AI autonomy with human oversight is crucial to fully realizing its capabilities. For more information on FPT.AI solutions, contact us via email: support@fpt.ai or hotline: 1900 638 399.

>>> EXPLORE:

• What Are Intelligent Agents? The Difference Between AI Agents and Intelligent Agents
• What is a Multi Agent System (MAS)?

PRIVACY POLICY 

(Code: 03-2026/CS/PDP Version: 2.0)  

1. Introduction

FPT Smart Cloud Company Limited (“FPT Smart Cloud”, “We”, “Us”) is committed to respecting and protecting the privacy and personal data of our customers, partners, and users throughout the process of interacting with, accessing, and using our products, services, platforms, and websites. This Privacy Statement (“Statement”) is designed to provide you with a comprehensive, transparent, and systematic view of how we collect, process, use, store, share, and protect personal data. 

This Statement also helps you understand your rights regarding personal data in accordance with the law. It is issued and implemented in compliance with current Vietnamese legal regulations, including but not limited to Personal Data Protection Law No. 91/2025/QH15, Decree 356/2025/ND-CP, and related guiding documents, while referencing international data protection practices and standards. 

2. Scope of Application

This Privacy Statement applies to all individuals interacting with FPT Smart Cloud, including but not limited to: 

• Visitors and users of the website; 
• Customers using products and services; 
• Business partners; 
• Individuals contacting, exchanging, or interacting with us. 

The scope covers all interaction channels, including websites, digital service platforms, applications, and other online communication methods operated or managed by FPT Smart Cloud. 

3. Personal Data Processing Roles

Depending on the specific relationship with you and the type of service provided, FPT Smart Cloud may act as: 

• Personal Data Controller: When we decide the purposes and means of data processing; 
• Personal Data Processor: When we process data at the request of customers or partners; 
• Both Controller and Processor of personal data. 

Specific roles will be determined based on the nature of the service and the agreement between the parties. 

4. Types of Personal Data Collected

We may collect and process the following types of personal data: 

4.1. Data Provided by you includes information you actively provide, such as: 

• Full name; 
• Email address; 
• Phone number; 
• Organization/Business information; 
• Content of exchanges, requests, or feedback. 

4.2. Automatically Collected Data Technical information collected when you access our website or services: 

• IP address; 
• Device type, operating system, and browser; 
• Access behavior data; 
• Cookies and similar tracking technologies. 

4.3. Data Arising During Service Use  

• System logs; 
• Usage history; 
• Interaction information with the platform. 

5. Purposes of Processing Personal Data

Personal data is processed for specific, legal, and transparent purposes, including: 

• Providing, operating, maintaining, and improving the quality of products and services; 
• Establishing, performing, and managing contractual relationships; 
• Receiving and responding to requests, inquiries, or complaints; 
• Ensuring system safety, security, and fraud prevention; 
• Fulfilling legal obligations as required by law; 
• Analyzing, researching, and enhancing user experience; 
• Conducting marketing and communication activities (subject to your consent). 

Note: We commit not to use your personal data to train Artificial Intelligence (AI) models without valid consent or unless the data has been anonymized according to regulations. 

6. Legal Basis for Processing

The processing of your personal data is based on one or more of the following: 

• Your explicit and valid consent; 
• Contractual performance obligations between you and FPT Smart Cloud; 
• Legal obligations we must comply with; 
• Legitimate interests of FPT Smart Cloud, provided they do not affect your legal rights and interests. 

7. Sharing and Disclosure of Personal Data

When necessary, your data may be shared with: 

• Affiliates and member units within the FPT ecosystem; 
• Partners and service providers supporting operations (e.g., technology infrastructure, technical support); 
• Competent state authorities as required by law; 
• Other relevant parties to protect the legal rights and interests of FPT Smart Cloud. 

We ensure that all data recipients are obligated to maintain confidentiality and only process data for the specified purposes. 

8. International Data Transfer

Personal data may be transferred outside the territory of Vietnam if necessary for service provision. In such cases, we commit to: 

• Fully complying with Vietnamese legal regulations on overseas data transfer; 
• Performing impact assessments and necessary legal procedures; 
• Applying appropriate protection measures to ensure data safety; 
• Fulfilling reporting obligations to competent authorities as required. 

9. Data Retention Period

Your personal data will be stored for the period necessary to fulfill the identified processing purposes or as required by applicable law. Upon expiration of the storage period, data will be deleted, anonymized, or destroyed using appropriate technical and organizational measures. 

10. Personal Data Security

We apply appropriate technical and organizational security measures to protect personal data from risks such as unauthorized access, loss, disclosure, or alteration. These include: 

• Access control; 
• Data encryption; 
• Intrusion monitoring and detection; 
• Security awareness training for personnel. 

11. Handling Personal Data Incidents

In the event of a personal data incident, FPT Smart Cloud will proactively take measures to control and rectify the situation. We will: 

• Notify competent authorities as required by law; 
• Notify the data subject when necessary to minimize risks. 

12. Your Rights

You have full rights related to your personal data as prescribed by law, including the right to be informed, access, correction, deletion, restriction or objection to processing, withdrawal of consent, and the right to complain. Detailed procedures are available in our Data Subject Rights Performance Policy published on our website. 

13. Cookies and Tracking Technologies

We use cookies to ensure website operation, analyze user behavior, personalize experiences, and support marketing activities (with your consent). Detailed usage is governed by our Cookie Policy. 

14. Third-Party Links

Our website may contain links to third-party websites or services. We do not control and are not responsible for the content or data protection policies of these parties. You should consult their privacy policies before providing personal data. 

15. Contact Information

For any questions, requests, or complaints regarding personal data processing, please contact: 

• Data Protection Officer (DPO): Pham The Minh 
• Company: FPT Smart Cloud Company Limited  
• Address: No 10 Pham Van Bach Street, Cau Giay ward, Hanoi, Vietnam 
• Email: support@fpt.ai  
• Hotline: 1900 638 399  

16. Statement Updates

FPT Smart Cloud reserves the right to modify or update this Privacy Statement periodically to ensure compliance with legal regulations and operational practices. Updated versions will be published on the website and take effect from the time of posting. Continued use of our services after an update signifies that you have read, understood, and agreed to the revised content. 

 

COOKIE POLICY

1. Introduction

This Cookie Policy (“Policy”) explains how FPT Smart Cloud Company Limited (“FPT Smart Cloud”, “We”, “Us”) uses cookies and similar technologies when you access and use our websites, platforms, and online services. 

This Policy should be read in conjunction with our Personal Data Protection Policy. 

Your continued use of the website after being notified about cookies implies your consent to the use of cookies in accordance with this Policy, unless you choose to opt out or adjust your cookie settings.

2. What are Cookies?

Cookies are small data files stored on your device (computer, phone, tablet) when you visit a website. Cookies allow the website to: 

• Recognize your device; 
• Remember your preferences and browsing behavior; 
• Improve performance and user experience. 

3. Purpose of Using Cookies

We use cookies for the following purposes: 

• Ensuring the stable operation of the website; 
• Remembering user choices and preferences; 
• Analyzing traffic and usage behavior; 
• Improving content, performance, and experience; 
• Supporting marketing activities (subject to your consent); 
• Ensuring security and detecting fraud. 

4. Classification of Cookies

We use the following types of cookies: 

4.1. Necessary Cookies These are essential for the website to function correctly. They handle: 

• Page navigation; 
• User authentication; 
• System security. These cookies do not require your consent. 

4.2. Performance and Analytics Cookies 

These cookies help us: 

• Understand how users interact with the website; 
• Collect aggregate information (non-directly identifiable); 
• Improve system performance. The use of these cookies requires your consent according to legal regulations. 

4.3. Functional Cookies 

These cookies allow the website to: 

• Remember your preferences (language, region); 
• Personalize the user experience. 

4.4. Marketing Cookies 

These cookies are used to: 

• Display relevant advertising content; 
• Measure the effectiveness of marketing campaigns; 
• Limit the frequency of advertisement displays. These cookies are only used with your consent. 

5. Third-Party Cookies

In some cases, we may use cookies from third parties, such as: 

• Analytics providers; 
• Advertising platforms; 
• Integrated services (video, maps, social networks). 

These third parties may collect and process data according to their own policies. We recommend that you consult the privacy policies of the relevant third parties. 

6. Managing and Setting Cookies

You have the right to choose whether to accept or decline cookies. You can manage cookies by: 

Adjusting browser settings: 

• Deleting cookies; 
• Blocking cookies; 
• Setting alerts when a cookie is sent. 

Using the cookie management tool on the website (if available): 

• Accepting all; 
• Declining; 
• Customizing by cookie type. 

Note: Declining certain types of cookies may affect your experience using the website. 

7. Legal Basis

Our use of cookies complies with: 

• Personal Data Protection Law No. 91/2025/QH15 and Decree 356/2025/ND-CP guiding the Personal Data Protection Law; 
• The principles of transparency and data subject consent; 
• International practices regarding privacy. 

For non-essential cookies, we will: 

• Only use them with your valid consent; 
• Allow you to withdraw your consent at any time. 

8. Cookie Storage Duration

Cookies can be stored: 

• In-session (session cookies): Automatically deleted when you close your browser. 
• Persistent cookies: Stored for a specific period of time. 

The specific storage time depends on the purpose of use and system configuration. 

9. Policy Updates

We may update this Cookie Policy from time to time to align with: 

• Changes in legislation; 
• Technological changes; 
• Operational needs. 

The updated version will be published on the website and will take effect from the time of posting. 

10. Contact

If you have any questions regarding the use of cookies, please contact: 

• Data Protection Officer (DPO)  
• Company: FPT Smart Cloud Company Limited  
• Address: No 10 Pham Van Bach Street, Cau Giay District, Hanoi, Vietnam  
• Email: Minhpt@fpt.com  
• Hotline: 1900638399  

PERSONAL DATA PROTECTION POLICY 

(Reference No.: 01-2026/CS/PDP – Version 2.0) 

1. Introduction

FPT Smart Cloud Company Limited (“FPT Smart Cloud”, “We”, “Us”) is committed to respecting and protecting the privacy and personal data of individuals and organizations when accessing and using our websites, platforms, and services. 

This Personal Data Protection Policy (“Policy”) is issued to: 

• Ensure transparency in personal data processing activities; 
• Safeguard the lawful rights and interests of Personal data subject matters; 
• Comply with Vietnamese laws, including the Law on Personal Data Protection No. 91/2025/QH15, Decree No. 356/2025/ND-CP, and relevant implementing regulations; 
• Align with international data protection standards. 

By accessing and using our platforms, you acknowledge that you have read, understood, and agreed to this Policy. 

2. Scope of Application

This Policy applies to all activities involving the collection and processing of personal data of: 

• Customers; 
• Website users; 
• Partners; 
• Individuals interacting or having contact with FPT Smart Cloud. 

This Policy applies across all channels, including but not limited to: 

• Websites; 
• Applications; 
• Digital service platforms; 
• Other online interaction channels. 

3. Definitions

For the purposes of this Policy: 

• Personal data: Personal data refers to digital data or information in other forms that identifies or assists the identification of a specific individual, including basic personal data and sensitive personal data. Personal data, once de-identified, is no longer considered personal data. 
• Personal data subject matters (or Data subject): Personal data subject matters refer to persons reflected in the personal data. 
• Personal data processing: Personal data processing refers to activities impacting personal data, including one or more of the following: collection, analysis, summary, encryption, decryption, modification, deletion, destruction, de-identification, provision, disclosure, transfer of personal data, and other activities impacting personal data.. 
• Personal data controlling party (or Data Controller): Personal data controlling party refers to an agency, organization, or individual that decides on the purposes and means of personal data processing. 
• Personal data processing party (or Data Processor): Personal data processing party refers to an agency, organization, or individual processing personal data as requested by the personal data processing party or personal data processing and controlling party under a contract. 
• Personal data processing and controlling party (or Data Controller and Processor): Personal data processing and controlling party refers to an agency, organization, or individual that decides on the purposes and means of personal data processing and directly processes personal data. 

Depending on specific circumstances, FPT Smart Cloud may act as a Personal data controlling party, Personal data processing party, or both Personal data processing and controlling party. 

4. Categories of Personaldatacollected 

4.1. Data provided by you 

Including but not limited to: 

• Full name; 
• Email address; 
• Phone number; 
• Company information; 
• Content of communications and requests. 

4.2. Automatically collected data 

When you access our website: 

• IP address; 
• Device and browser type; 
• Cookie data; 
• Browsing behavior information. 

4.3. Data generated during service use 

• System logs; 
• Usage history; 
• Platform interaction data. 

5. Purposes ofdataprocessing 

Personal data is processed for the following purposes: 

• Providing, operating, and improving services; 
• Establishing, performing, and managing contractual relationships; 
• Receiving and handling requests and feedback; 
• Ensuring safety and information security; 
• Complying with legal obligations; 
• Conducting analysis and research to enhance service quality; 
• Carrying out marketing and communication activities with consent. 

FPT Smart Cloud commits not to use customer personal data to train artificial intelligence (AI) models without explicit consent or full anonymization. 

6. Legalbasis forprocessing 

We process personal data based on: 

• Consent of the Personal data subject matter; 
• Contractual obligations; 
• Legal obligations under applicable laws; 
• Legitimate interests of FPT Smart Cloud, provided such interests do not override the lawful rights and interests of the Personal data subject matter. 

7. Datasharing anddisclosure 

Personal data may be shared in the following cases: 

• With affiliated companies and subsidiaries; 
• With partners and service providers (e.g., cloud infrastructure providers, technical support); 
• Upon request by competent state authorities; 
• When necessary to protect the lawful rights and interests of FPT Smart Cloud. 

Recipients of data are obligated to maintain confidentiality and process data only for agreed purposes. 

8. Cross-borderdatatransfer 

As a principle, FPT Smart Cloud stores data within Vietnam. In cases where cross-border data transfer is required: 

• We comply with Vietnamese legal regulations on data transfer; 
• We conduct a Data Transfer Impact Assessment and submit the required dossier to the Department of Cybersecurity and High-Tech Crime Prevention (A05) under the Ministry of Public Security; 
• We apply appropriate safeguards to ensure data security. 

9. Dataretentionperiod 

Personal data is retained for as long as necessary to fulfill the processing purposes or as required by law. 

After the retention period, data will be: 

• Deleted; 
• Anonymized; or 
• Securely destroyed in accordance with internal procedures. 

10. Rights ofPersonal data subject matters

Personal data subject matters have rights under applicable laws, including: 

• The right to be informed and to give consent; 
• The right to access and correct data; 
• The right to withdraw consent; 
• The right to request deletion of data; 
• The right to restrict or object to processing; 
• The right to lodge complaints or initiate legal action. 

To exercise these rights, Personal data subject matters may contact us using the details provided in Section 13. 

FPT Smart Cloud will handle such requests within the timeframes required by applicable law. 

11. Personaldatasecurity 

We implement appropriate technical and organizational measures to protect personal data, including: 

• Access control mechanisms; 
• Data encryption; 
• System monitoring; 
• Prevention of unauthorized access; 
• Staff training on information security. 

12. Personaldatabreach handling 

In the event of a personal data incident, FPT Smart Cloud will: 

• Promptly detect and assess the incident; 
• Apply necessary remedial measures; 
• Notify competent authorities within 72 hours (or as required by law); 
• Notify affected Personal data subject matters when necessary. 

13. Contactinformation

For any requests or inquiries regarding personal data, please contact: 

Data Protection Officer (DPO) 
FPT Smart Cloud Company Limited 

Address: No. 10 Pham Van Bach Street, Cau Giay Ward, Hanoi, Vietnam 

Email: Minhpt@fpt.com 

Phone: 0913571357 

Hotline: 1900638399 

14. Implementation Provisions

This Policy may be amended or supplemented from time to time to ensure compliance with legal requirements and operational practices. 

This Policy takes effect from the date it is officially published on the FPT Smart Cloud website. 

SECURITY OR PRIVACY VULNERABILITY REPORTING 

If you believe you have discovered a security vulnerability or a privacy-related issue within FPT Smart Cloud’s products or services, we look forward to receiving your feedback to help build a safer cloud computing environment. 

I. Reporting Methods

FPT Smart Cloud welcomes reports from security researchers, partners, and customers. To ensure information is processed quickly, please send an email to the following addresses: 

• Technical Security Issues: support@fpt.ai
• Privacy & Personal Data Issues: minhpt@fpt.com (Send directly to the Data Protection Officer). 

Your report should include the following information: 

• The affected product/service and software version. 
• A detailed description of the observed behavior versus the expected behavior. 
• Specific steps to reproduce the error (include illustrative videos if available). 
• Safety Note: Please encrypt sensitive information before sending it to protect the data during transmission. 

II. Receipt and Processing Procedure

To protect our systems and customer data, FPT Smart Cloud applies the Coordinated Vulnerability Disclosure (CVD) process according to ISO/IEC 29147 standards: 

• Acknowledgment: We will send a confirmation response acknowledging receipt of your report within 24-48 business hours. 
• Investigation & Remediation: FPT Smart Cloud will not disclose or publicly discuss security issues until the investigation is complete and necessary updates have been widely released. 
• Public Disclosure: We use Security Advisories to provide information about fixes and to recognize the contributions of the reporting individuals or organizations. 

III. Security Advisories & Compliance 

FPT Smart Cloud’s Security Advisories are designed to help customers maintain system safety and meet standards such as PCI DSS v4.0.1 and ISO 27001. 

• Technical Advisories: Updates on configuration changes, patches, or potential risks that are not categorized as software vulnerabilities but affect overall information security. 
• Recommendations: Customers should regularly monitor Technical Security Advisories and update their systems periodically to ensure data availability and integrity. 

FPT Smart Cloud is committed to maintaining the confidentiality of the reporter’s information and handling all vulnerabilities with the highest sense of responsibility. 

Artificial Intelligence (AI) is reshaping our world at a rapid pace. From optimizing business processes to supporting critical decisions, AI offers infinite potential, fostering a synergy between AI and humans. However, alongside its immense benefits, AI also poses significant challenges, especially concerning ethics. The question is no longer “How will AI develop?” but rather “How should AI develop responsibly?”. 

This is where AI ethics becomes a timely and urgent topic, not only for technology developers but also for regulators, businesses, and society as a whole. Let’s explore what AI ethics is with FPT.AI! 

What is AI Ethics? 

AI ethics is a set of principles, values, and guidelines that aim to ensure the transparent, fair, and responsible development, deployment, and use of artificial intelligence. According to UNESCO, AI ethics doesn’t just focus on what the technology can do, but also revolves around the questions of “What should AI do?” and “How should we use AI to benefit humanity, rather than cause harm?”. 



Why is AI Ethics Important?  

AI is no longer merely supporting technology. It’s a powerful tool capable of making decisions. According to Deputy Minister of Science and Technology, Mr. Bui The Duy, AI is completely different from any other technologies that humans have found before. While old technological products only follow existing instructions, AI can create its own directions, beyond the control of developers. 

Furthermore, AI has now become an indispensable partner for humans. A global study involving approximately 32,000 workers from 47 countries by The University of Melbourne showed that over 58% of employees actively use AI in their work, with one-third of them using AI weekly or daily. . 



Therefore, without ethical oversight, sometimes wrong decisions made by AI can have far-reaching impacts on individuals or even a nation. An algorithm could reject a job application simply because their resume came from a rural area, implicitly assuming they are “less promising.” A facial recognition system could misidentify people of color due to a lack of diverse data. A chatbot could learn discriminatory language from social media users if left unchecked. 

These examples highlight that if AI lacks ethics, the consequences will not be limited to technical errors but will also lead to social, legal, and humanitarian consequences. 

Core Values Shaping AI for the Future 



In the journey of artificial intelligence development, what’s important is not just how far technology advances, but which direction we are leading it. To ensure AI serves the common good, for people, society, and this planet, UNESCO has outlined four core values that act as guiding principles, including: 

• Respecting human rights and human dignity: Ensuring the respect, protection, and promotion of human rights, fundamental freedoms, and the dignity of each individual. 
• Building peaceful, just, and interconnected societies: Encouraging the development of societies where everyone can live in harmony, fairness, and connection. 
• Promoting diversity and inclusion: AI must be designed to serve everyone, excluding no one, fostering diversity and creating equal opportunities. 
• Protecting the environment and developing thriving ecosystems: AI technology needs to be environmentally responsible, contributing to the protection of the planet and natural ecosystems. 

These values are crucial compasses for guiding AI development in a positive, sustainable direction. 

Core Principles in AI Ethics

With the goal of “living safely” with AI, philosopher Luciano Floridi has distilled 5 core principles, becoming reliable beacons on the journey alongside AI: 

• Beneficence: AI should be developed and applied to improve the lives of humans and our planet. The ultimate goal is to create “AI for Social Good” (AI4SG), where AI is used to enhance societal well-being. 
• Nonmaleficence: The principle of “do no harm” is paramount, especially when AI has the potential to affect human existence. This includes avoiding harm to privacy, autonomy, and employment opportunities. 
• Autonomy: Human ability to act freely and independently must be preserved and promoted, while machine autonomy needs to be limited. 
• Justice: AI must be developed, designed, and deployed in a way that promotes justice, fairness, equality, and related values. This requires addressing issues like algorithmic bias and ensuring equitable access. 
• Explicability: To promote other principles, we need to understand the “how” and “why” behind AI systems and products. Accountability and transparency are key. 

What Should Businesses Do? 

For businesses, investing in AI technology cannot be separated from building an ethical foundation. According to Coursera, many global corporations like IBM, Google, and Microsoft have established internal ethics committees, built codes of conduct, and verification processes to ensure their AI products adhere to ethical standards from the outset. 

Some specific actions regarding AI ethics that businesses can take include: 

• Training personnel on AI ethics, data, and privacy. 
• Integrating ethical risk assessment into the product development process. 
• Consulting independent experts to review critical algorithms. 
• Being transparent about how data is collected, processed, and used. 
• Establishing internal ethics councils to verify products before launch. 

Not Just One Person’s Responsibility 

AI ethics is not solely a matter for the tech industry; it’s a shared responsibility of society as a whole: 

• Nations and government agencies need to create flexible, appropriate, and practical legal frameworks that keep pace with technological development while still safeguarding human rights. 
• Universities and research institutes need to integrate ethics topics into AI and data science curricula. 
• AI users also need to enhance their understanding to use AI intelligently and responsibly. 

Towards a Sustainable Technological Future 

AI ethics is not a barrier to innovation, but a solid foundation for the development of artificial intelligence. By prioritizing these core values, we are building a future where AI is not only exceptionally intelligent but also deeply humane, serving and enhancing human life. 

As AI becomes increasingly integrated into decision-making systems, AI ethics is no longer an option – it’s a prerequisite for building a just, transparent, and humane digital society. 

🤝 Build ethical AI today, create lasting trust tomorrow. 

Sources

UNESCO. (n.d.). Recommendation on the Ethics of Artificial Intelligence. https://www.unesco.org/en/artificial-intelligence/recommendation-ethics 

Coursera. (2023, July 26). What is AI ethics? Definition and examples. https://www.coursera.org/articles/ai-ethics 

IBM. (n.d.). AI ethics. IBM. https://www.ibm.com/think/topics/ai-ethics 

University of Texas at Austin. (n.d.). AI ethics. Ethics Unwrapped. https://ethicsunwrapped.utexas.edu/glossary/ai-ethics thics Unwrapped. https://ethicsunwrapped.utexas.edu/glossary/ai-ethics