Legal

AI & Security Policy

iFortis Worldwide Private Limited
AI Native Enterprise Transformation · Intelligence as a Infrastructure™

Effective Date: 9 August 2026
Last Updated: 9 August 2026

1. Purpose

Artificial intelligence is becoming foundational to how enterprises operate, create, decide and transform.

At iFortis Worldwide Private Limited (“iFortis Worldwide”, “iFortis”, “we”, “us” or “our”), we believe enterprise AI must be built with the same discipline expected of critical enterprise technology.

This AI & Security Policy establishes the principles and controls that guide how iFortis designs, develops, deploys, operates and governs AI-enabled platforms, software, digital products and enterprise transformation solutions.

Our objective is to build AI that is:

Secure. Responsible. Governed. Resilient. Transparent. Enterprise-ready.

This Policy applies alongside our Privacy Notice, Terms of Use, applicable Data Processing Agreements, Enterprise Agreements, Statements of Work, security requirements and other contractual commitments.

2. Scope

This Policy applies to AI systems and technology environments operated, developed or managed by iFortis, including where applicable:

  • AI platforms;
  • generative AI systems;
  • machine-learning systems;
  • intelligent software;
  • AI agents;
  • autonomous or semi-autonomous workflows;
  • enterprise AI solutions;
  • AI-enabled digital products;
  • AI engineering environments;
  • data and analytics platforms;
  • AI integrations;
  • model infrastructure;
  • third-party AI services; and
  • AI capabilities incorporated into enterprise transformation engagements.

Where iFortis provides services to an enterprise customer, specific contractual requirements may supplement or supersede this Policy.

3. Our AI Governance Principles

Our approach to AI is built around eight principles.

3.1 Human Accountability

AI is designed to augment human intelligence and enterprise capabilities.

Where decisions may have material consequences for individuals, organisations or society, appropriate human oversight should remain in place.

AI should not become an excuse for removing accountability.

3.2 Security by Design

Security is considered from architecture through deployment and ongoing operation.

Security controls may include:

  • identity and access management;
  • authentication;
  • authorisation;
  • encryption;
  • secure development;
  • network controls;
  • vulnerability management;
  • monitoring;
  • logging;
  • incident response; and
  • controlled infrastructure access.

Security is treated as an engineering responsibility, not a compliance activity.

3.3 Privacy by Design

AI systems may process significant volumes of information.

We therefore seek to incorporate privacy considerations into:

Data → Architecture → Model → Application → Deployment → Monitoring

Personal information should be processed only for appropriate and lawful purposes and in accordance with applicable privacy requirements and contractual obligations.

3.4 Responsible Data Use

Data used within AI systems should be relevant to the intended purpose and handled according to applicable requirements.

Depending on the engagement, controls may address:

  • data classification;
  • data minimisation;
  • access restrictions;
  • retention;
  • deletion;
  • data lineage;
  • provenance;
  • consent or lawful basis;
  • customer instructions; and
  • contractual restrictions.
3.5 Transparency

Where reasonably appropriate, users and enterprise customers should understand:

  • when AI is being used;
  • what the system is designed to do;
  • relevant limitations;
  • whether outputs are generated or retrieved;
  • where human review is expected; and
  • how AI-related risks are managed.

The precise level of transparency depends on the product, use case and applicable legal requirements.

3.6 Reliability and Evaluation

AI systems may behave differently from conventional deterministic software.

Accordingly, appropriate systems may be evaluated for:

  • accuracy;
  • consistency;
  • robustness;
  • security;
  • reliability;
  • harmful or unexpected behaviour;
  • performance;
  • hallucination or fabrication risk;
  • prompt manipulation;
  • model misuse; and
  • operational resilience.

Evaluation requirements are determined according to the intended use and risk profile.

3.7 Fairness and Non-Discrimination

Where AI systems may affect individuals or groups, we seek to identify and reduce unreasonable risks of discriminatory or unfair outcomes.

Depending on the use case, this may include consideration of:

  • representative data;
  • model behaviour;
  • evaluation criteria;
  • human review;
  • output monitoring; and
  • escalation mechanisms.

AI systems should not be intentionally designed to discriminate unlawfully.

3.8 Accountability

Every AI system should have an appropriate level of ownership and operational accountability.

Depending on the system, this may involve:

  • product owners;
  • engineering teams;
  • security teams;
  • data owners;
  • enterprise stakeholders;
  • governance functions; and
  • authorised decision-makers.

4. AI Risk Classification

Not every AI system presents the same level of risk.

iFortis may classify AI use cases according to factors including:

  • purpose;
  • users;
  • autonomy;
  • data sensitivity;
  • potential impact;
  • regulatory requirements;
  • operational criticality;
  • security exposure; and
  • potential consequences of failure.

Illustrative risk levels may include:

Low Risk

AI used for:

  • general productivity;
  • content assistance;
  • summarisation;
  • non-sensitive recommendations; or
  • internal experimentation.
Moderate Risk

AI supporting:

  • enterprise workflows;
  • business analysis;
  • customer interactions;
  • operational recommendations;
  • document processing; or
  • decision-support systems.
High Risk

AI that may materially influence:

  • employment;
  • access to important services;
  • financial outcomes;
  • health;
  • safety;
  • legal rights;
  • security;
  • critical infrastructure; or
  • other high-impact decisions.

Higher-risk use cases may require additional controls, validation, human oversight, monitoring and approval before deployment.

5. AI Lifecycle Governance

AI governance begins before deployment and continues throughout the system lifecycle.

Our lifecycle may include:

Discover → Assess → Design → Build → Test → Approve → Deploy → Monitor → Improve → Retire

At relevant stages, teams may assess:

  • intended purpose;
  • data sources;
  • security requirements;
  • privacy implications;
  • model behaviour;
  • potential misuse;
  • operational risks;
  • human oversight;
  • regulatory requirements; and
  • exit or retirement requirements.

6. Model Governance

Where iFortis develops, configures or operates AI models, appropriate governance may include:

  • model identification;
  • model versioning;
  • configuration management;
  • evaluation;
  • testing;
  • performance monitoring;
  • change management;
  • access controls;
  • documentation;
  • incident management; and
  • retirement procedures.

Where third-party foundation models or AI services are used, their limitations and contractual conditions may also be considered.

7. Generative AI Governance

Generative AI may generate text, code, images, recommendations, structured information or other outputs.

Because generated output may be probabilistic, users should not assume that every output is:

  • accurate;
  • complete;
  • original;
  • current;
  • legally appropriate; or
  • suitable for a particular business purpose.

Appropriate validation should be applied before consequential use.

8. AI Agents and Autonomous Systems

AI agents may be capable of:

  • planning;
  • reasoning;
  • retrieving information;
  • invoking tools;
  • interacting with systems;
  • executing workflows; or
  • taking actions within authorised environments.

Where agentic functionality is deployed, appropriate controls may include:

  • scoped permissions;
  • tool restrictions;
  • authentication;
  • action approval;
  • execution boundaries;
  • logging;
  • monitoring;
  • rate limits;
  • human intervention;
  • rollback mechanisms; and
  • emergency shutdown or suspension procedures where technically applicable.

AI agents should not receive broader permissions than necessary for their intended function.

9. Prompt and Input Security

AI systems may be exposed to malicious or unintended instructions.

Depending on the architecture, iFortis may implement controls addressing risks such as:

  • prompt injection;
  • indirect prompt injection;
  • data exfiltration;
  • malicious instructions;
  • context manipulation;
  • unsafe tool execution;
  • privilege escalation; and
  • model abuse.

Input validation, access controls and system-level safeguards may be applied according to the relevant risk.

10. Output Security

AI-generated outputs may introduce security, privacy or operational risks.

Where appropriate, systems may incorporate controls such as:

  • output filtering;
  • policy enforcement;
  • content validation;
  • sensitive-data detection;
  • access controls;
  • human review;
  • monitoring; and
  • escalation.

Customers remain responsible for validating outputs before using them in consequential business processes unless otherwise agreed.

11. Enterprise Customer Data

Enterprise customer data is handled according to:

  • the applicable enterprise agreement;
  • data-processing terms;
  • customer instructions;
  • product architecture;
  • applicable law; and
  • agreed security requirements.

Where iFortis acts as a processor or service provider, we process customer data within the scope authorised by the applicable agreement.

Enterprise data does not automatically become part of a general-purpose AI training dataset.

Any permitted secondary use, model improvement activity or other processing will be determined by the applicable contractual and privacy framework.

12. Data Segregation

Where appropriate to the product architecture, controls may be used to separate:

  • customer environments;
  • customer data;
  • production and development environments;
  • administrative access;
  • testing environments; and
  • internal corporate systems.

Specific isolation mechanisms depend on the applicable deployment architecture.

13. Identity and Access Management

Access to systems and data should follow the principle of least privilege.

Depending on the environment, this may include:

  • role-based access control;
  • authentication;
  • privileged-access management;
  • access logging;
  • credential protection;
  • environment restrictions;
  • administrative controls; and
  • access review.

Access should be granted based on legitimate business or operational requirements.

14. Encryption

Where appropriate, data may be encrypted:

  • during transmission;
  • while stored; and/or
  • within applicable processing environments.

Specific encryption mechanisms may vary according to:

  • product;
  • infrastructure;
  • deployment model;
  • customer requirements; and
  • technical architecture.

Security-sensitive technical details may be made available to qualified enterprise customers through appropriate security-review processes.

15. Secure Software Development

Security considerations may be incorporated throughout the software-development lifecycle.

Depending on the product, activities may include:

  • secure architecture;
  • code review;
  • dependency management;
  • vulnerability assessment;
  • security testing;
  • access-control testing;
  • secrets management;
  • environment protection;
  • logging; and
  • remediation.

Security requirements may increase according to the criticality and risk profile of the system.

16. Vulnerability Management

iFortis seeks to identify and address vulnerabilities affecting its technology environment.

Vulnerabilities may be identified through:

  • internal testing;
  • security monitoring;
  • customer reports;
  • responsible disclosure;
  • assessments;
  • automated tools; or
  • other security activities.

Vulnerabilities may be prioritised according to factors including:

  • severity;
  • exploitability;
  • exposure;
  • affected assets;
  • business impact; and
  • potential customer impact.

17. Third-Party AI Providers

iFortis may use third-party:

  • AI models;
  • foundation models;
  • cloud infrastructure;
  • APIs;
  • software;
  • security services;
  • data services; and
  • other technology providers.

Before incorporating third-party technology into relevant enterprise solutions, iFortis may consider factors such as:

  • security;
  • privacy;
  • reliability;
  • contractual restrictions;
  • data handling;
  • availability;
  • technical suitability;
  • regulatory considerations; and
  • business continuity.

Applicable subprocessors may be disclosed through appropriate contractual or enterprise documentation.

18. AI Supply-Chain Security

AI systems may depend upon complex technology supply chains.

Where relevant, iFortis may consider risks associated with:

  • third-party models;
  • software dependencies;
  • libraries;
  • APIs;
  • datasets;
  • infrastructure providers;
  • model updates;
  • external integrations; and
  • service dependencies.

Changes to critical dependencies may be subject to appropriate technical and operational review.

19. Monitoring and Logging

Where appropriate, systems may generate logs relating to:

  • authentication;
  • access;
  • administrative actions;
  • system events;
  • security events;
  • AI interactions;
  • tool execution;
  • errors;
  • performance; and
  • operational activity.

Logging is used for legitimate purposes such as security, troubleshooting, reliability, auditing and incident response, subject to applicable privacy and contractual requirements.

20. Security Incident Management

iFortis maintains processes intended to detect, assess, contain, investigate and respond to security incidents.

Response activities may include:

Detect → Assess → Contain → Investigate → Remediate → Recover → Review

Where notification is required by applicable law or contract, iFortis will follow the applicable notification requirements.

Enterprise-specific incident response obligations may be established through contractual arrangements.

21. Business Continuity and Resilience

Depending on the relevant service, iFortis may maintain controls supporting:

  • backup;
  • recovery;
  • redundancy;
  • monitoring;
  • incident response;
  • disaster recovery; and
  • operational continuity.

Specific service availability, recovery objectives, RTOs, RPOs and service credits are governed by the applicable enterprise agreement or SLA where applicable.

22. AI Misuse Prevention

iFortis seeks to prevent its technology from being intentionally used for unlawful or harmful purposes.

Customers and users must not use iFortis systems to:

  • conduct unlawful activity;
  • compromise systems;
  • facilitate fraud;
  • create malicious software;
  • bypass security controls;
  • obtain unauthorised information;
  • conduct harassment or abuse;
  • impersonate individuals unlawfully;
  • manipulate systems without authorisation;
  • violate privacy rights; or
  • otherwise misuse AI capabilities.

Additional restrictions may apply to specific products.

23. Human Review and High-Impact Decisions

AI output should not automatically determine high-impact outcomes.

Where AI is used in contexts involving significant effects on individuals, appropriate human oversight should be established.

Examples may include:

  • employment;
  • credit;
  • insurance;
  • healthcare;
  • education;
  • legal matters;
  • safety;
  • access to essential services; and
  • other materially consequential decisions.

Applicable laws may impose additional requirements.

24. AI Documentation

Depending on the system and risk profile, relevant documentation may include:

  • system purpose;
  • intended use;
  • prohibited use;
  • data sources;
  • model information;
  • known limitations;
  • evaluation results;
  • security considerations;
  • deployment architecture;
  • human oversight requirements;
  • monitoring requirements; and
  • incident procedures.

The level of documentation will be proportionate to the system and its intended use.

25. Employee and Contractor Responsibilities

Personnel who develop, deploy, administer or operate AI systems are expected to:

  • follow applicable security requirements;
  • protect confidential information;
  • use authorised systems;
  • maintain appropriate access controls;
  • report security concerns;
  • follow AI governance procedures;
  • avoid unauthorised data use; and
  • complete applicable training.

Violations may result in appropriate disciplinary or contractual action.

26. Customer Responsibilities

Enterprise customers are responsible for:

  • configuring authorised users appropriately;
  • protecting credentials;
  • determining lawful use cases;
  • obtaining required permissions;
  • validating AI outputs where appropriate;
  • complying with applicable law;
  • protecting information submitted to the platform;
  • maintaining appropriate human oversight; and
  • following agreed security procedures.

Customers should not provide information to an iFortis system that they are not legally or contractually authorised to provide.

27. Security and AI Assessments

Qualified enterprise customers may request relevant security, privacy or AI-governance information for procurement and risk-assessment purposes.

Subject to confidentiality, security and commercial considerations, iFortis may provide appropriate documentation relating to:

  • security architecture;
  • data flows;
  • AI governance;
  • subprocessors;
  • data processing;
  • access controls;
  • business continuity;
  • incident response;
  • security practices; and
  • applicable assurance information.

Certain sensitive technical information may be restricted to protect the security of iFortis systems and its customers.

28. Certifications and Assurance

Where iFortis holds specific security, privacy, AI or compliance certifications or independent assurance reports, applicable information may be made available through the appropriate enterprise or trust channel.

Nothing in this Policy should be interpreted as representing that iFortis holds a certification, accreditation or assurance report unless expressly confirmed by iFortis.

29. Policy Exceptions

Certain enterprise engagements may require additional controls or contractual arrangements.

Exceptions to this Policy should be:

  • documented;
  • risk assessed;
  • approved by the appropriate authority; and
  • subject to applicable contractual and legal requirements.

30. Policy Review

This Policy may be reviewed and updated periodically in response to:

  • technology developments;
  • emerging AI risks;
  • security threats;
  • regulatory requirements;
  • customer requirements;
  • changes to our products; and
  • lessons learned from operational experience.

The latest version will be published through the appropriate iFortis channel.

31. Responsible Disclosure

Security researchers, customers and other parties who identify a potential vulnerability are encouraged to report it responsibly.

Security & AI Governance Contact

business@ifortisworldwide.org

Please provide:

  • affected product or system;
  • description of the issue;
  • reproduction steps;
  • potential impact;
  • relevant evidence; and
  • contact information.

Please do not publicly disclose vulnerabilities before iFortis has had a reasonable opportunity to investigate and address them.

32. Contact

iFortis Worldwide Private Limited

Level 3, Commercial Development, Manhattan Street
6 Worldmark, Asset, LP-1B-03, Aerocity
New Delhi, Delhi 110037
India

AI, Security & Enterprise Trust

business@ifortisworldwide.org

Our Position

AI should not merely be intelligent.

It should be responsible by design, secure by architecture and accountable in operation.

At iFortis Worldwide®, we build AI Native enterprise transformation around that principle.

AI Native Enterprise Transformation
Intelligence as a Infrastructure™

Intelligence engineered for enterprise.

Contact us Submit an RFQ