AI Contracts: Key Clauses Every Business Should Include in AI Agreements
Quick Answer: An AI contract should clearly address the AI system being supplied, permitted uses, data processing, intellectual-property rights, ownership or licensing of outputs, confidentiality, cybersecurity, model updates, performance standards, warranties, indemnities, limitation of liability, audit rights, subcontractors, regulatory compliance and termination. The exact clauses depend on whether the agreement concerns an AI SaaS product, API, model licence, development project, procurement arrangement or AI-enabled service.
A company signs a contract with an AI provider.
The provider promises access to an advanced AI system.
The company uploads customer data.
Employees use the system to generate business documents.
The AI provider subsequently changes the model.
The output quality declines.
A customer alleges that one of the generated outputs infringes its copyright.
A data-protection incident occurs.
Who is responsible?
The answer may depend heavily on the contract.
AI agreements are therefore becoming an important part of technology-law practice.
Unlike conventional software contracts, AI agreements may need to address additional questions:
- How is the AI model trained?
- Can customer data be used for model improvement?
- Who owns AI-generated outputs?
- Can the provider change the model?
- What happens if model performance deteriorates?
- What happens if the AI produces harmful or unlawful outputs?
- Who is responsible for regulatory compliance?
- What happens when the relationship ends?
These questions should ideally be answered before the AI system is deployed.
This article explains the principal clauses businesses should consider when negotiating AI agreements.
Legal disclaimer: This article provides general educational information and is not legal advice. AI contracting requirements vary according to jurisdiction, industry, transaction structure and the nature of the AI system.
Key Takeaways
- AI contracts should identify the exact AI system, services and permitted use cases.
- Businesses should clearly define who owns or licenses AI outputs.
- Training data and customer data should be governed expressly rather than indirectly.
- Contracts should state whether customer inputs can be used to train or improve models.
- Data-protection roles should be documented where personal data is processed.
- Confidentiality clauses should address AI-specific risks.
- Cybersecurity obligations should be specific rather than limited to general promises of “reasonable security”.
- AI performance should be linked to measurable service levels where appropriate.
- Contracts should address model updates, retraining and material changes.
- Indemnification should be negotiated carefully, particularly for intellectual-property claims.
- Liability caps should be aligned with the risks and value of the AI deployment.
- Audit and documentation rights can become important when AI systems are complex or opaque.
- Subprocessors and other third parties should be identified and controlled.
- Businesses should establish clear exit, deletion and transition procedures.
Why Are AI Contracts Different From Traditional Software Contracts?
Quick Answer: AI systems can be probabilistic, continuously updated, data-dependent and capable of producing outputs that neither party can fully predict in advance.
Traditional software usually performs predefined functions.
AI systems can behave differently depending on:
- Input data.
- Model version.
- Prompt design.
- Training data.
- System configuration.
- Fine-tuning.
- Safety controls.
This creates contractual uncertainty.
A customer may purchase access to an AI system today and receive materially different outputs after a model update six months later.
The contract should therefore address not only the initial system but also the lifecycle of the AI service.
What Is an AI Contract?
Quick Answer: An AI contract is an agreement governing the development, licensing, supply, deployment, integration or use of an artificial-intelligence system or service.
AI contracts can take many forms.
- AI SaaS agreements.
- AI API agreements.
- Model licences.
- AI development agreements.
- AI procurement agreements.
- AI consulting agreements.
- Data-processing agreements.
- AI-powered software licences.
- Research and development agreements.
- Enterprise AI master service agreements.
Clause 1: Define the AI Services
Quick Answer: The contract should clearly define the AI product, model, services, features and permitted functionality being supplied.
Ambiguous descriptions can create disputes.
A contract stating only:
“Provider shall provide access to its AI platform.”
may be insufficient.
The agreement should identify, where appropriate:
- Model name.
- Model version.
- API services.
- Available features.
- Usage limits.
- Supported integrations.
- Deployment environment.
- Geographic availability.
Clause 2: Define Permitted AI Use
Quick Answer: The contract should identify how the customer may use the AI system.
Permitted use may include:
- Internal business operations.
- Customer support.
- Content generation.
- Software development.
- Research.
- Data analysis.
Restrictions should also be clear.
For example, a provider may prohibit use in:
- High-risk applications.
- Medical diagnosis.
- Financial decisions.
- Employment decisions.
- Safety-critical systems.
The contract should identify these restrictions before implementation.
Clause 3: AI Data Ownership
Quick Answer: AI contracts should clearly establish ownership and permitted use of data supplied to the AI system.
Businesses should distinguish between:
- Customer data.
- Provider data.
- Training data.
- Input data.
- Output data.
- Usage data.
- Telemetry.
These categories should not be treated as interchangeable.
Clause 4: Can the Provider Use Customer Data to Train AI?
Quick Answer: This should be expressly addressed in the contract rather than left to assumption.
Businesses should determine whether customer inputs can be:
- Stored.
- Reviewed by humans.
- Used for model improvement.
- Used for training.
- Shared with affiliates.
- Shared with subprocessors.
Where personal data is involved, the contractual arrangement must also be assessed under applicable data-protection law.
The ICO specifically recommends that organisations assess the controller/processor relationship throughout the AI supply chain and document the relationship in contracts or other agreements. ([ico.org.uk](https://ico.org.uk/for-organisations/advice-and-services/audits/data-protection-audit-framework/toolkits/artificial-intelligence/contracts-and-third-parties/?utm_source=chatgpt.com))
Clause 5: Intellectual Property Ownership
Quick Answer: AI contracts should separately address ownership of pre-existing intellectual property, AI models, customer inputs, outputs and human-created modifications.
Businesses should identify:
- Background IP.
- Provider IP.
- Customer IP.
- Model IP.
- Training materials.
- Customer inputs.
- AI outputs.
- Derived materials.
This distinction prevents disputes over ownership.
Who Owns AI-Generated Outputs?
Quick Answer: The contract should state the commercial rights granted to the customer, but contractual language cannot necessarily create intellectual-property rights that do not exist under applicable law.
This is particularly important because copyrightability of AI-generated material can depend on human authorship.
As discussed in Article #54, businesses should distinguish between contractual control over outputs and statutory copyright ownership.
A contract might provide that the customer may use outputs commercially.
That does not automatically mean the customer owns copyright in every element of the output.
Clause 6: AI Output Rights
Quick Answer: The contract should specify whether the customer receives ownership, a licence or another contractual right to use AI outputs.
The agreement should address:
- Commercial use.
- Internal use.
- Modification.
- Redistribution.
- Sublicensing.
- Publication.
- Resale.
The contract should also identify whether the provider makes any representations concerning third-party rights in outputs.
Clause 7: Intellectual-Property Indemnity
Quick Answer: An IP indemnity can allocate certain risks arising from claims that the AI service or its outputs infringe third-party intellectual-property rights.
Businesses should examine:
- What claims are covered?
- Does the indemnity cover outputs?
- Does it cover the underlying model?
- Are customer modifications excluded?
- Are prohibited uses excluded?
- Is there a monetary cap?
- Who controls the defence?
An AI provider may offer a broad-looking indemnity containing significant exclusions.
Those exclusions should be reviewed carefully.
Clause 8: Data Protection
Quick Answer: Where an AI provider processes personal data, the contract should clearly establish the parties' data-protection roles and obligations.
Depending on the arrangement, parties may need to determine whether they are:
- Controller and processor.
- Joint controllers.
- Independent controllers.
The ICO recommends documenting controller and processor relationships across the AI supply chain. ([ico.org.uk](https://ico.org.uk/for-organisations/advice-and-services/audits/data-protection-audit-framework/toolkits/artificial-intelligence/contracts-and-third-parties/?utm_source=chatgpt.com))
Clause 9: Data Processing Agreement
Quick Answer: Where the legal framework requires a controller-processor arrangement, an appropriate data-processing agreement should establish the processor's obligations.
Under Article 28 of the UK GDPR, processor contracts must address matters including the subject matter and duration of processing, nature and purpose, types of personal data, categories of data subjects, documented instructions, confidentiality, security, subprocessors, assistance with data-subject rights and end-of-contract provisions. ([ico.org.uk](https://ico.org.uk/for-organisations/uk-gdpr-guidance-and-resources/accountability-and-governance/contracts-and-liabilities-between-controllers-and-processors-multi/what-needs-to-be-included-in-the-contract/?q=rights&utm_source=chatgpt.com))
AI contracts should therefore be reviewed alongside the applicable data-processing agreement rather than treating privacy as a separate afterthought.
Clause 10: Confidentiality
Quick Answer: AI contracts should clearly protect confidential information supplied to the AI system.
Confidentiality provisions should address:
- Business information.
- Trade secrets.
- Source code.
- Customer information.
- Product plans.
- Research.
- Financial information.
The contract should also specify who within the provider organisation may access customer information.
Clause 11: AI Cybersecurity
Quick Answer: AI contracts should contain specific security obligations appropriate to the system's risk.
Possible requirements include:
- Encryption.
- Access controls.
- Authentication.
- Vulnerability management.
- Security testing.
- Incident response.
- Logging.
- Backup.
- Business continuity.
The ICO's AI contracting guidance recommends specifying technical and organisational security measures, including encryption, pseudonymisation or minimisation, system resilience and backup arrangements. ([ico.org.uk](https://ico.org.uk/for-organisations/advice-and-services/audits/data-protection-audit-framework/toolkits/artificial-intelligence/contracts-and-third-parties/?utm_source=chatgpt.com))
Clause 12: Security Incident Notification
Quick Answer: AI contracts should establish how quickly and through what process the provider must notify the customer of security incidents.
The agreement should specify:
- Notification deadline.
- Notification method.
- Information to be provided.
- Investigation obligations.
- Remediation.
- Customer cooperation.
Shorter notification periods may be appropriate for high-risk AI systems.
Clause 13: AI Performance and Service Levels
Quick Answer: AI contracts should establish measurable performance standards where the customer depends on predictable system performance.
Possible metrics include:
- Availability.
- Response time.
- Latency.
- Accuracy.
- Error rates.
- Throughput.
- Support response time.
The appropriate metric depends on the AI system.
The ICO specifically recommends considering accuracy-based KPIs or service-level agreements when procuring AI systems or datasets. ([ico.org.uk](https://ico.org.uk/for-organisations/advice-and-services/audits/data-protection-audit-framework/toolkits/artificial-intelligence/contracts-and-third-parties/?utm_source=chatgpt.com))
Clause 14: Accuracy and AI Performance Warranties
Quick Answer: AI warranties should be carefully drafted because AI outputs are often probabilistic and cannot realistically be guaranteed to be accurate in every case.
Instead of promising:
“The AI will always produce accurate results.”
the contract may establish measurable performance standards appropriate to the use case.
Businesses should also determine how performance will be measured and what happens when agreed thresholds are not met.
Clause 15: Model Updates and Changes
Quick Answer: Contracts should explain when and how an AI provider can modify, replace, retrain or upgrade the model.
This is one of the most important AI-specific clauses.
Businesses should consider:
- Notice before material changes.
- Testing requirements.
- Customer approval.
- Backward compatibility.
- Performance testing.
- Security review.
- Regulatory impact.
- Termination rights.
A provider should not necessarily be able to make unlimited changes to a mission-critical AI system without contractual consequences.
Clause 16: Continuous Learning
Quick Answer: Where an AI system continuously learns or changes based on new data, the contract should establish how those changes are controlled and documented.
This is particularly important for high-risk or regulated applications.
The contract should identify:
- What data can modify the system.
- Who approves retraining.
- How changes are tested.
- How changes are documented.
- Whether the customer receives notice.
- What happens if performance deteriorates.
Clause 17: Audit Rights
Quick Answer: Audit rights allow customers to assess whether AI providers are complying with contractual, security, privacy and performance obligations.
Audit provisions can cover:
- Security.
- Data protection.
- AI governance.
- Performance.
- Subprocessors.
- Regulatory compliance.
For complex AI systems, customers may also seek independent audit or certification evidence.
Clause 18: Documentation Rights
Quick Answer: AI contracts should specify what technical and compliance documentation the provider must maintain and make available.
Possible documentation includes:
- Model specifications.
- Model cards.
- Training documentation.
- Testing results.
- Risk assessments.
- Security documentation.
- Version histories.
- Incident records.
The EU AI Act requires providers of general-purpose AI models to maintain technical documentation and provide information and documentation to downstream AI-system providers so that they can understand model capabilities and limitations and comply with their own obligations. ([digital-strategy.ec.europa.eu](https://digital-strategy.ec.europa.eu/en/faqs/guidelines-obligations-general-purpose-ai-providers?utm_source=chatgpt.com))
Clause 19: Regulatory Compliance
Quick Answer: AI contracts should allocate responsibility for compliance with applicable AI, privacy, cybersecurity, consumer-protection and sector-specific laws.
The agreement should address:
- Applicable AI legislation.
- Data-protection law.
- Consumer law.
- Sector regulation.
- Export controls where relevant.
- Cybersecurity requirements.
The parties should avoid simply stating:
“Each party shall comply with all applicable laws.”
That clause may be necessary but can be insufficient.
Responsibility should be allocated according to the actual activities of each party.
Clause 20: EU AI Act Responsibilities
Quick Answer: Contracts involving AI systems within the EU should identify which party is acting as provider, deployer or another relevant actor and allocate the operational responsibilities associated with that role.
The EU AI Act uses a risk-based regulatory structure.
For general-purpose AI models, providers have obligations including technical documentation, copyright policies and training-content summaries. ([digital-strategy.ec.europa.eu](https://digital-strategy.ec.europa.eu/en/factpages/general-purpose-ai-obligations-under-ai-act?utm_source=chatgpt.com))
The European Commission's current guidance also states that the GPAI obligations have applied since 2 August 2025, while Commission enforcement powers for those obligations entered into application on 2 August 2026. ([digital-strategy.ec.europa.eu](https://digital-strategy.ec.europa.eu/en/policies/guidelines-gpai-providers?utm_source=chatgpt.com))
Contracting parties should therefore ensure that their agreements reflect the actual regulatory roles and responsibilities involved.
Clause 21: AI Intellectual-Property Indemnification
Quick Answer: IP indemnities can be particularly important where AI outputs or underlying AI technology may trigger third-party intellectual-property claims.
The contract should clarify whether indemnification covers:
- Copyright claims.
- Trademark claims.
- Patent claims.
- Trade-secret claims.
- Model infringement.
- AI-generated outputs.
Customers should carefully examine exclusions for:
- Customer prompts.
- Customer data.
- Customer modifications.
- Unapproved uses.
Clause 22: Limitation of Liability
Quick Answer: AI contracts should establish appropriate financial limits for liability while identifying risks that may require higher caps or separate treatment.
A standard contract might contain a general liability cap.
But AI deployments may involve risks that are disproportionate to the contract value.
Businesses should therefore consider separate treatment for:
- Confidentiality breaches.
- Data-protection violations.
- Cybersecurity incidents.
- IP infringement.
- Fraud.
- Wilful misconduct.
- Personal injury.
The enforceability of liability limitations depends on applicable law.
Clause 23: Indemnification
Quick Answer: Indemnification clauses determine when one party must compensate another for specified third-party claims, losses or liabilities.
AI contracts should define:
- Covered claims.
- Procedural requirements.
- Defence control.
- Settlement authority.
- Excluded circumstances.
- Financial limits.
Clause 24: Human Oversight
Quick Answer: Contracts for AI systems affecting people should address human oversight where appropriate.
The agreement may establish:
- Who reviews AI decisions.
- When human intervention is mandatory.
- Escalation procedures.
- Override mechanisms.
- Training requirements.
This is especially relevant to recruitment, finance, healthcare, insurance and other high-impact applications.
Clause 25: Subprocessors and Third Parties
Quick Answer: AI providers frequently rely on cloud providers, model providers, data providers and other subcontractors, so contracts should address the wider technology supply chain.
The ICO recommends written authorisation and contractual controls for subprocessors where personal data is processed. ([ico.org.uk](https://ico.org.uk/for-organisations/uk-gdpr-guidance-and-resources/accountability-and-governance/contracts-and-liabilities-between-controllers-and-processors-multi/when-is-a-contract-needed-and-why-is-it-important/?utm_source=chatgpt.com))
Businesses should consider:
- Approved subprocessors.
- Change notification.
- Objection rights.
- Flow-down obligations.
- Security requirements.
- Data location.
Clause 26: Geographic Restrictions
Quick Answer: Contracts should identify where AI services and associated data may be processed.
This can matter for:
- Data protection.
- Data localisation.
- Government restrictions.
- Sector regulations.
- Export controls.
Cloud architecture can make geographic processing more complicated than it appears from the provider's headquarters.
Clause 27: Data Retention and Deletion
Quick Answer: AI contracts should establish how long customer information, prompts, outputs and logs are retained and what happens when the agreement ends.
The contract should address:
- Retention periods.
- Deletion requests.
- Backup copies.
- Training datasets.
- Logs.
- Archived information.
The ICO recommends contractual provisions requiring processors to delete or return personal information at the end of the contract unless law requires continued storage. ([ico.org.uk](https://ico.org.uk/for-organisations/advice-and-services/audits/data-protection-audit-framework/toolkits/artificial-intelligence/contracts-and-third-parties/?utm_source=chatgpt.com))
Clause 28: Termination Rights
Quick Answer: AI contracts should identify circumstances allowing termination, particularly where the system becomes materially unsuitable, unlawful or unsafe.
Termination triggers can include:
- Material breach.
- Repeated SLA failure.
- Security incidents.
- Regulatory changes.
- Loss of required certification.
- Material model changes.
- Inability to meet compliance requirements.
Clause 29: Exit and Transition Assistance
Quick Answer: Businesses should negotiate transition rights before becoming dependent on an AI provider.
Exit provisions can cover:
- Data export.
- Model or configuration migration.
- Documentation transfer.
- Deletion certification.
- Transition support.
- Continued access for a limited period.
This is particularly important where switching AI providers would be technically difficult.
Clause 30: Business Continuity
Quick Answer: Critical AI contracts should address what happens if the provider suffers an outage, cyberattack, insolvency event or other major disruption.
Possible requirements include:
- Disaster recovery.
- Backups.
- Redundancy.
- Recovery time objectives.
- Recovery point objectives.
- Incident communications.
AI Contract Clause Checklist
| Clause | Key Question |
|---|---|
| Services | What exactly is being supplied? |
| Permitted use | How can the AI be used? |
| Data | Who owns and controls the data? |
| Training | Can customer data train the model? |
| IP | Who owns or licenses the relevant IP? |
| Outputs | What rights does the customer receive? |
| Privacy | Who is controller and processor? |
| Security | What security controls apply? |
| Performance | What happens if performance declines? |
| Updates | Can the provider change the model? |
| Indemnity | Who bears third-party claims? |
| Liability | What are the financial limits? |
| Audit | Can compliance be independently assessed? |
| Termination | When can the contract be ended? |
| Exit | How can the customer migrate away? |
AI Contract Due-Diligence Checklist
Quick Answer: Contract review should begin before negotiation by examining the AI provider and the technology itself.
- Identify the AI provider.
- Identify the underlying model.
- Identify all subprocessors.
- Determine where data is processed.
- Review the provider's security controls.
- Review the provider's AI documentation.
- Determine whether customer data is used for training.
- Review output rights.
- Assess IP indemnification.
- Assess liability caps.
- Assess regulatory compliance.
- Review termination provisions.
- Review data deletion provisions.
- Review business continuity arrangements.
What Should a Business Never Leave Ambiguous in an AI Contract?
Quick Answer: Businesses should avoid ambiguity around data use, output rights, model changes, liability, security, privacy and termination.
At minimum, the following questions should have clear answers:
- Can the provider train on our data?
- Who owns our inputs?
- Who can use our outputs?
- What happens to our data after termination?
- Can the provider materially change the model?
- Who pays if the AI infringes third-party IP?
- Who is responsible for a data breach?
- What happens if the AI becomes unavailable?
- What happens if the provider becomes insolvent?
- Can we audit compliance?
Frequently Asked Questions
What is an AI contract?
An AI contract is an agreement governing the development, licensing, procurement, deployment, integration or use of an artificial-intelligence system or service.
What clauses should an AI contract contain?
Important clauses can cover AI services, permitted use, data ownership, training data, intellectual property, outputs, privacy, security, warranties, model updates, indemnification, liability, audit rights, subcontractors and termination.
Who owns AI-generated content under an AI contract?
The contract can establish contractual rights concerning outputs, but statutory copyright ownership depends on applicable law and the nature of human creative contribution.
Can an AI provider use customer data to train its model?
That depends on the agreement and applicable law. Businesses should ensure that training or model-improvement rights are expressly addressed rather than assumed.
Should AI contracts contain an IP indemnity?
For many commercial AI deployments, an IP indemnity can be important, particularly where the customer faces potential third-party claims concerning the AI service or outputs.
What is an AI liability clause?
An AI liability clause allocates financial responsibility for specified losses or claims arising from the AI service, subject to applicable law and negotiated limitations.
Should AI contracts include a liability cap?
Usually, commercial AI agreements address liability caps, but the appropriate cap depends on the value and risk of the deployment. Some categories of liability may require separate treatment under applicable law.
Should AI contracts address model updates?
Yes. Contracts should explain when providers can modify, replace, retrain or materially change AI models and what rights the customer has when those changes materially affect the service.
Can AI providers change their models without notice?
Whether they can depends on the contract. Customers should negotiate appropriate notice and protection against material changes where model consistency is commercially important.
What should an AI data-processing agreement cover?
Where a controller-processor relationship exists, the agreement should address processing purposes, data categories, instructions, confidentiality, security, subprocessors, data-subject rights, assistance, deletion and other legally required matters.
Should AI vendors provide audit rights?
Audit or assurance rights can be important for high-risk AI systems, particularly where customers need to verify security, privacy, performance or regulatory compliance.
What happens to AI data when the contract ends?
The contract should establish whether data is returned, deleted or retained under a defined legal basis and retention period.
Should AI contracts identify subprocessors?
Yes. Businesses should understand which third parties process their data or support the AI service and establish appropriate contractual controls.
Are AI contracts different from SaaS contracts?
AI contracts often contain many traditional SaaS provisions but may require additional terms concerning model training, outputs, probabilistic performance, model changes, AI governance and intellectual-property risks.
Do AI contracts need to address the EU AI Act?
Where the EU AI Act applies, contracts should help allocate the operational responsibilities of the relevant AI actors and support compliance with applicable obligations.
What should businesses do before signing an AI agreement?
Businesses should conduct technical, legal, privacy, security and commercial due diligence and negotiate the allocation of risks before deploying the AI system.
Conclusion
An AI contract is no longer simply a software licence with the word “AI” added to it.
Artificial intelligence creates a different contractual environment.
Models can change.
Outputs can be unpredictable.
Training data can create intellectual-property questions.
Customer information can become model input.
AI systems can depend on multiple third-party providers.
And regulatory responsibilities can be distributed across the technology supply chain.
This makes contractual risk allocation increasingly important.
A well-drafted AI agreement should answer several fundamental questions.
What is the customer buying?
What data can the provider use?
Can customer data be used to train models?
Who controls the outputs?
What happens when the model changes?
Who is responsible for security?
Who bears intellectual-property risk?
What happens when the system fails?
How can the customer exit?
These questions should be resolved before the AI system becomes embedded in business operations.
The ICO's current AI contracting guidance emphasises the importance of assessing the controller-processor relationship across the supply chain, conducting appropriate due diligence before procurement and establishing clear contractual responsibilities. ([ico.org.uk](https://ico.org.uk/for-organisations/advice-and-services/audits/data-protection-audit-framework/toolkits/artificial-intelligence/contracts-and-third-parties/?utm_source=chatgpt.com))
For general-purpose AI, the EU AI Act also creates documentation and information obligations for providers, including information that must be made available to downstream AI-system providers. ([digital-strategy.ec.europa.eu](https://digital-strategy.ec.europa.eu/en/faqs/guidelines-obligations-general-purpose-ai-providers?utm_source=chatgpt.com))
That means AI contracts increasingly operate as part of an organisation's broader AI-governance framework.
The contract should therefore be connected to:
- AI risk assessments.
- Data-protection assessments.
- Cybersecurity requirements.
- Procurement procedures.
- Intellectual-property policies.
- Incident-response plans.
- Business-continuity arrangements.
Businesses should also avoid relying entirely on standard vendor terms.
A generic SaaS agreement may not adequately address model training, AI outputs, continuous learning, AI-specific indemnities or regulatory responsibilities.
The strongest AI contracts do not simply describe the technology. They allocate responsibility for the risks created by the technology.
As AI becomes embedded into ordinary commercial operations, AI contracting will increasingly become a core component of technology procurement and legal risk management.
Legal Disclaimer
This article is provided for general educational and informational purposes only. It is not legal, contractual, technology, intellectual-property, data-protection or regulatory advice and does not create an attorney-client relationship. AI contracts should be reviewed by qualified counsel familiar with the applicable jurisdiction, industry and technology.