15. Emerging Technologies and Future Security Strategy

Introduction and Context

Enterprise security is increasingly shaped by technologies that alter not only the tools used by organisations, but also the location, timing and meaning of trust. Cloud platforms distribute computation across regions and providers; the Internet of Things extends enterprise systems into homes, factories, vehicles and public infrastructure; and artificial intelligence enables machines to make security decisions at operational speed. At the same time, quantum computing threatens some widely used cryptographic assumptions, while 5G and future 6G networks create highly programmable and decentralised communication environments.

These developments cannot be managed through technology acquisition alone. They require a security strategy that is anticipatory, risk-based, ethically defensible and adaptable. A security leader must ask not merely whether a technology is innovative, but what trust assumptions it changes, which new dependencies it creates, how failures will be contained, and whether its benefits justify its environmental, social and operational costs.

This chapter examines confidential computing, post-quantum cryptography, advanced mobile networks, edge artificial intelligence, autonomous security, digital twins and decentralised technologies. It then integrates them into a strategic approach for enterprise ecosystems. The central argument is that future security will depend on combining technological controls with governance, human judgement, resilience engineering and continuous reassessment.

Learning Objectives

After studying this chapter, students should be able to:

1. Explain the security purpose and operating principles of major emerging technologies.

2. Critically compare their benefits, attack surfaces, dependencies and limitations.

3. Assess how confidential computing and post-quantum cryptography change enterprise trust models.

4. Analyse the security implications of 5G/6G, edge AI, digital twins and decentralised architectures.

5. Design a phased, ethical and sustainable security roadmap for a future enterprise ecosystem.

6. Evaluate the roles of governance, automation, human oversight, privacy and resilience in technology adoption.

From Perimeter Defence to Distributed Trust

Traditional enterprise security assumed that valuable information was located within a relatively controlled network. Security architecture therefore concentrated on firewalls, central identity directories and inspection at a network boundary. Cloud, mobile and IoT environments have weakened this assumption. Devices may be unmanaged, workloads may be ephemeral, users may operate from any location, and data may be processed by several organisations.

The resulting security model is commonly associated with zero trust: access is continuously evaluated using identity, device state, workload context, policy and observed behaviour. Emerging technologies extend this model by introducing new forms of hardware- and software-based trust. A confidential computing platform may protect data while it is being processed. A cryptographic migration programme may preserve trust against future quantum attacks. A digital twin may provide a model of physical operations, while edge AI may make local decisions without sending all data to a central cloud.

However, distributed trust is not equivalent to universal trust. Each component introduces assumptions about firmware, vendors, algorithms, data quality, availability and governance. A sound strategy makes these assumptions explicit and tests them against failure scenarios.

Confidential Computing: Protecting Data During Use

Encryption traditionally protects data at rest and in transit. Confidential computing addresses a third state: data in use. It uses hardware-supported trusted execution environments (TEEs) to isolate selected code and data from the operating system, hypervisor, cloud administrator or other workloads. The protected region is often called an enclave.

A typical confidential-computing process involves four stages. First, a workload is placed inside a protected execution environment. Second, the environment is measured, producing evidence about the code and configuration running inside it. Third, a remote party performs attestation, checking whether that evidence satisfies a policy. Finally, sensitive keys or data are released only after successful attestation.

This model is valuable in multi-party analytics, financial services, healthcare, government workloads and cloud services in which organisations need to collaborate without exposing raw information. For example, several hospitals might calculate population-level statistics while restricting access to individual patient records.

Confidential computing does not eliminate all risk. A TEE can contain vulnerabilities in its processor, firmware, enclave interface or attestation service. Side-channel attacks may infer information from timing, memory access or resource use. An authorised application inside an enclave may still misuse data. Confidentiality also does not guarantee correctness, availability or lawful processing. Consequently, TEEs should complement, rather than replace, encryption, secure coding, access governance, monitoring and data minimisation.

Strategically, organisations should identify workloads where the threat model includes privileged infrastructure administrators, cloud co-tenants or cross-organisational data exposure. They should then evaluate the assurance evidence, vendor dependence, performance overhead, recovery process and exit options before adopting the technology.

Post-Quantum Cryptography and Cryptographic Agility

Quantum computers capable of breaking widely deployed public-key systems are not assumed to be available for ordinary enterprise use today, but the risk is strategic because sensitive information may be collected now and decrypted later. This is often described as a harvest-now, decrypt-later threat. Long-lived information—such as identity records, research, diplomatic material and industrial designs—requires early protection.

Post-quantum cryptography (PQC) consists of cryptographic algorithms designed to resist attacks by both conventional and quantum computers. It is distinct from quantum key distribution, which depends on specialised physical communication systems. PQC is generally intended to operate through software and existing digital networks, although algorithm integration may affect performance, certificate size and protocol design.

A migration programme should begin with cryptographic inventory. Organisations need to discover where public-key algorithms are used in certificates, VPNs, application protocols, code-signing systems, identity platforms, databases and embedded devices. They must classify information according to confidentiality lifetime and identify suppliers whose products cannot yet support algorithm replacement.

Cryptographic agility is the ability to change algorithms, key sizes and protocols without redesigning the entire system. It requires abstraction in software, algorithm negotiation with secure downgrade prevention, automated certificate management, tested key rotation and documented dependencies. A premature replacement of all algorithms may cause interoperability and operational problems; delaying all action may leave high-value information exposed. A risk-based transition commonly combines current strong cryptography with approved PQC mechanisms in hybrid configurations while standards and implementations mature.

PQC itself has limitations. New algorithms may have larger keys or signatures, implementation vulnerabilities and uncertain performance in constrained IoT devices. Migration also creates supply-chain and configuration risks. The appropriate objective is not to predict the exact date of quantum capability, but to reduce irreversible dependence on cryptographic assumptions that cannot be changed quickly.

5G, 6G and the Expansion of the Attack Surface

5G networks introduce high bandwidth, low latency, network slicing, virtualised network functions and extensive use of software-defined infrastructure. These capabilities support industrial automation, connected vehicles, remote operations and massive IoT deployments. They also shift security from a relatively specialised telecom environment toward cloud-native software, APIs, orchestration platforms and third-party service ecosystems.

Network slicing allows logical networks with different performance and security requirements to share physical infrastructure. Isolation failures, insecure slice orchestration or compromised management interfaces could therefore affect multiple services. Virtualised functions may be exposed to vulnerabilities familiar from cloud environments, including insecure images, weak identity controls and misconfigured APIs. The radio access network introduces additional concerns involving device authentication, rogue components, signalling abuse and availability.

6G research is expected to extend these trends through more integrated sensing and communications, highly distributed architectures, advanced automation and closer coupling with AI. Because technical specifications and deployment models remain under development, security strategy should focus on durable principles rather than speculative features: strong supply-chain assurance, identity-based access, segmentation, secure update mechanisms, observability, resilience and independent testing.

Enterprises using private 5G should treat the network as an enterprise platform rather than merely a connectivity service. Responsibilities must be allocated among the mobile operator, equipment provider, cloud provider, integrator and customer. Contracts should address vulnerability disclosure, software bills of materials, incident notification, logging access, lawful data handling and service continuity.

Edge AI and Security Decision-Making at the Point of Data

Edge AI places machine-learning inference near sensors, devices or local gateways rather than sending all data to a central cloud. Its advantages include lower latency, reduced bandwidth use, improved operational continuity and, in some cases, greater privacy. Examples include anomaly detection in manufacturing, local medical-device analysis and visual inspection in logistics.

The edge also changes the threat model. Devices may be physically accessible and computationally constrained. Attackers can attempt model extraction, evasion, poisoning, firmware compromise or manipulation of sensor inputs. Data drift can cause a model to perform poorly as equipment, users or environments change. A model may also encode unfair or unsafe decisions, particularly where training data do not represent operating conditions.

Security for edge AI requires a lifecycle approach. Devices should be provisioned with unique identities, secure boot and signed firmware. Models should be signed, versioned and deployed through controlled processes. Sensitive telemetry should be minimised and protected. Local decisions should include confidence thresholds and safe fallback modes. High-impact actions should be subject to human approval or independent safety controls. Monitoring must cover both conventional security events and model behaviour, including drift, unusual confidence patterns and distribution changes.

An important design principle is graceful degradation. If a local model loses connectivity to its management service, encounters uncertain inputs or detects tampering, it should fail into a safe and operationally understood state—not silently continue making high-consequence decisions.

Autonomous Security: Speed with Accountability

Autonomous security uses automation, machine learning and orchestration to detect, investigate and respond to threats. Examples include automated isolation of a device, dynamic adjustment of access policy, prioritisation of vulnerabilities and generation of incident hypotheses. The value of autonomy is greatest where attack speed exceeds human response capacity and where repetitive decisions can be standardised.

Yet autonomy creates a delegation problem. A response system may block a critical production service, delete evidence, quarantine a legitimate user or amplify a false positive. Organisations must therefore distinguish between advisory, human-approved and fully automated actions. The level of autonomy should reflect reversibility, potential harm, confidence in the evidence and the criticality of the affected asset.

A useful governance model defines an action matrix. Low-impact and reversible actions, such as adding a temporary detection rule, may be automated. Medium-impact actions, such as isolating a workstation, may require an on-call approval unless predefined emergency conditions apply. Irreversible or safety-critical actions should require multiple controls and, where feasible, human authorisation.

Autonomous security systems need explainable records: input evidence, model or rule version, decision, action, authorisation path and rollback result. They should be tested against adversarial inputs and unexpected combinations of events. Automation must reduce cognitive burden rather than conceal uncertainty from analysts.

Digital Twins and the Security of Cyber-Physical Decisions

A digital twin is a continuously updated digital representation of a physical asset, process or environment. It may combine sensor data, engineering models, operational histories and simulation. In security, digital twins can support vulnerability analysis, incident rehearsal, predictive maintenance and safe testing of configuration changes.

For example, an organisation operating a water-treatment facility could model the effects of altered network rules or sensor anomalies before applying changes to the live system. A twin can help identify dependencies that are difficult to see in static asset inventories.

The security of a digital twin is inseparable from the security of its data and interfaces. If an attacker corrupts sensor feeds, the twin may produce an inaccurate representation and recommend unsafe action. If the model exposes operational details, it may become an intelligence resource for adversaries. Synchronisation delay, incomplete asset coverage and inaccurate assumptions can create false confidence.

Digital twins should therefore record data provenance, confidence and time. Simulations should be validated against observed behaviour. Access must be segmented according to engineering and security roles, and high-risk recommendations should be reviewed before execution. A twin is a decision-support instrument, not an unquestionable replica of reality.

Decentralised Technologies and New Trust Relationships

Decentralised technologies distribute data, control or validation across multiple participants rather than relying on one central authority. Distributed ledgers, decentralised identity, smart contracts and token-based systems are common examples. They may improve traceability, reduce dependence on a single intermediary and support multi-organisational processes.

Their security properties are frequently misunderstood. A distributed ledger may make recorded entries difficult to alter, but it does not establish that the original data were accurate. A smart contract may execute exactly as written while producing an undesirable result. Consensus mechanisms can create performance, energy, governance or concentration concerns. Publicly visible records may conflict with privacy, deletion and data-localisation requirements.

Decentralised identity can allow individuals or organisations to present verifiable credentials without a central identity store. This may reduce some breach risks, but revocation, recovery, key loss, wallet security and accessibility remain major issues. Enterprise adoption should focus on a clearly defined trust problem rather than treating decentralisation as an objective in itself.

A governance analysis should ask who can change protocols, resolve disputes, revoke credentials, recover from compromise and meet regulatory obligations. In many cases, a permissioned or hybrid design is more suitable than a fully public network. Privacy-enhancing techniques, selective disclosure and off-chain storage may reduce exposure, but they do not remove the need for lawful governance.

An Integrated Framework for Future Security Strategy

A future-oriented security roadmap should connect technology choices with business purpose, risk appetite and public responsibility. The following six-stage framework provides a practical method.

1. Establish strategic context

Map critical services, stakeholders, regulatory obligations, safety consequences and information lifetimes. Identify dependencies on cloud providers, telecom operators, chip manufacturers, AI platforms and software suppliers. The objective is to understand what must remain confidential, correct, available and recoverable.

2. Conduct a trust and threat assessment

For each proposed technology, document trust anchors, privileged parties, failure modes, attack paths and assumptions. Threat modelling should include malicious insiders, compromised suppliers, physical access, model manipulation, cryptographic compromise and outages. Use scenario analysis rather than relying only on compliance checklists.

3. Prioritise experiments with containment

Proofs of concept should use synthetic or minimised data, isolated environments and measurable success criteria. An experiment should test security assurance, interoperability, performance, operational skill requirements and exit feasibility. Innovation without containment can turn a pilot into an uncontrolled production dependency.

4. Build foundational capabilities

Before adopting advanced tools, strengthen identity governance, asset discovery, vulnerability management, secure software development, logging, backup, incident response and supplier assurance. Emerging technologies amplify weak foundations; they rarely compensate for them.

5. Scale through architecture and governance

Approved patterns should define encryption, key management, device identity, model deployment, attestation, segmentation, monitoring and human oversight. Boards and risk committees should receive measures that connect technical exposure with business impact, such as recovery time, critical supplier concentration, unsupported cryptographic assets and high-impact automated decisions.

6. Reassess continuously

Roadmaps should include review triggers: a major vulnerability, new cryptographic guidance, regulatory change, model drift, supplier acquisition, operational incident or change in threat capability. Security architecture must be treated as a living system rather than a one-time project.

A balanced roadmap often has three horizons. The immediate horizon inventories assets, cryptographic use, suppliers and high-risk automation. The medium horizon migrates priority systems, pilots confidential computing and edge controls, and exercises digital-twin and autonomous-response scenarios. The longer horizon addresses quantum-resistant interoperability, advanced network ecosystems, decentralised identity and new assurance methods. Timelines should be tied to risk and dependency maturity, not to promotional technology cycles.

Ethical, Sustainable and Regulatory Considerations

Future security decisions affect people who may not participate in them. Edge AI can monitor workers; autonomous systems can restrict access; decentralised records can make personal information difficult to remove; and digital twins can expose sensitive operational or environmental data. Ethical governance therefore requires necessity, proportionality, transparency, contestability and accountability.

High-impact automated decisions should have an appeal or review mechanism. Data collection should be limited to a legitimate purpose, and models should be evaluated for bias and unequal error rates. Security monitoring should not become indiscriminate surveillance. Responsibility must remain identifiable even when a decision is made by a distributed platform or autonomous agent.

Sustainability also belongs in security strategy. Energy-intensive computation, expanded sensor deployments, hardware replacement and duplicated data stores have environmental costs. Organisations should evaluate lifecycle impact, repairability, efficient model design, renewable-energy use, equipment reuse and secure disposal. A system that is technically secure but financially or environmentally unsustainable may not be resilient in the long term.

Regulatory requirements vary across jurisdictions, but recurring obligations include privacy protection, critical-infrastructure resilience, breach notification, algorithmic accountability, records management, supply-chain assurance and sector-specific safety. Legal review should occur during architecture design, not after deployment.

Applied Case Study: Northstar Manufacturing’s Adaptive Security Roadmap

Background

Northstar Manufacturing operates several plants, a cloud-based enterprise resource-planning system and a private 5G network connecting robots, sensors and maintenance devices. It plans to introduce edge AI for predictive maintenance and a digital twin for production-line optimisation. Its research division exchanges confidential designs with external partners through a cloud collaboration platform.

Situation and problem

A security review identified three concerns. First, plant devices used long-lived certificates and inconsistent update processes. Second, design files were processed by cloud services administered partly by external providers. Third, proposed autonomous controls could stop production when the AI detected abnormal behaviour, but no clear approval or rollback process existed. Senior management initially proposed adopting confidential computing, blockchain-based supplier records and autonomous response simultaneously.

Analysis

The security team separated business outcomes from technology preferences. The highest risks were compromise of plant availability, theft of designs and unsafe automated intervention. A dependency review showed that many devices could not immediately support larger post-quantum keys. The team also found that a ledger would not solve the accuracy problem in supplier data, since records entered by authorised users could still be false.

Approach and strategy

Northstar adopted a phased roadmap. It first established an inventory of devices, certificates, algorithms, data flows and suppliers. Plant networks were segmented, device identities were standardised and signed firmware updates were introduced. High-value design workloads were moved to a confidential-computing pilot using synthetic data, with attestation required before keys were released.

The organisation began cryptographic discovery and selected long-lived design information for early protection. It required vendors to publish algorithm-transition plans and tested hybrid protocol support in a non-production environment. Edge AI was deployed initially for advisory maintenance alerts rather than automatic shutdown. A human operator remained responsible for production decisions, while independent safety interlocks could stop equipment when physical thresholds were exceeded.

A limited digital twin was created for one production line. Its outputs included confidence and data freshness indicators, and proposed configuration changes required engineering approval. The organisation rejected a blockchain project because a conventional, governed supplier database met the traceability requirement more simply.

Outcome

The roadmap reduced immediate exposure without creating an unmanageable collection of experimental technologies. Northstar gained visibility into cryptographic dependencies, improved device accountability and tested confidential processing under controlled conditions. The edge-AI pilot demonstrated operational value while revealing model drift during seasonal production changes. This finding led to a formal model-monitoring and retraining process.

Learning points

The case illustrates that emerging-technology strategy should begin with mission risks and trust assumptions. Confidential computing addressed a specific cloud-provider threat; it was not treated as a universal solution. PQC preparation began with inventory and prioritisation rather than a disruptive replacement programme. Automation was introduced according to reversibility and safety impact. Finally, rejecting an unsuitable technology was itself a successful strategic decision.

Challenges and Common Strategic Errors

Several recurring mistakes weaken future security programmes. Organisations may adopt a technology because it is fashionable, without identifying a threat or measurable benefit. They may confuse decentralisation with security, assume that AI accuracy is stable, or regard attestation as proof that an entire service is trustworthy. Another error is to automate before improving asset data and response procedures; poor inputs then produce fast but unreliable decisions.

Vendor concentration is a further concern. A single cloud, chip, AI or telecom provider may become a critical dependency. Contracts, portability testing and alternative suppliers should be considered before the dependency becomes difficult to reverse. Skills shortages can also create hidden operational risk: an organisation may purchase advanced systems that it cannot independently assess, monitor or recover.

Finally, security teams sometimes measure activity rather than resilience. The number of AI alerts, encrypted workloads or migrated certificates is less significant than whether critical services can withstand compromise, recover within acceptable limits and preserve rights and safety.

Best Practices for Adaptable Enterprise Security

• Maintain an inventory of assets, algorithms, models, identities, suppliers and data flows.
• Make trust assumptions explicit and review them during architecture changes.
• Prefer secure, reversible experimentation over uncontrolled deployment.
• Use hardware roots of trust, signed updates, strong identity and least privilege for edge and IoT systems.
• Design cryptographic agility before quantum migration becomes urgent.
• Apply human oversight to high-impact or irreversible autonomous actions.
• Monitor model drift, data provenance and confidence, not only cyber events.
• Test failure, recovery and degraded-operation modes through exercises and simulation.
• Evaluate privacy, safety, environmental impact and accessibility alongside security.
• Tie investment decisions to business-critical outcomes, measurable risk reduction and exit options.

Chapter Summary

Emerging technologies are changing where computation occurs, how trust is established and how security decisions are made. Confidential computing protects data during processing but depends on trustworthy hardware, attestation and secure application design. Post-quantum cryptography requires early inventory, prioritisation and cryptographic agility because sensitive information may remain valuable for decades. 5G and 6G expand the enterprise attack surface through virtualisation, APIs, network slicing and complex supply chains. Edge AI improves latency and privacy but introduces model, device, sensor and lifecycle risks. Autonomous security can respond at machine speed only when actions are governed by confidence, reversibility and accountability. Digital twins support simulation and resilience but can mislead when their data or assumptions are corrupted. Decentralised technologies may redistribute trust, but they do not guarantee accurate data, privacy or effective governance.

An adaptable strategy begins with mission and risk context, inventories dependencies, conducts trust-based threat modelling, pilots technologies safely, strengthens foundational controls and reassesses continuously. Ethical governance, sustainability, regulatory compliance and human responsibility are not secondary concerns; they are conditions for durable security.

Key Terms

Confidential computing: Protection of data while it is being processed, usually through a trusted execution environment.

Trusted execution environment (TEE): A hardware-supported isolated area that protects code and data from unauthorised access.

Remote attestation: A process by which a system proves the measured state of its protected execution environment.

Post-quantum cryptography: Cryptographic methods designed to resist attacks by conventional and quantum computers.

Harvest-now, decrypt-later: The collection of encrypted information today for possible decryption when future capabilities improve.

Cryptographic agility: The ability to replace algorithms, keys or protocols with limited architectural disruption.

Network slicing: The creation of logically separated networks over shared physical 5G infrastructure.

Edge AI: Artificial intelligence inference performed near the data source rather than solely in a central cloud.

Model drift: Degradation in model performance caused by changes in data, environments or operating conditions.

Autonomous security: Automated or machine-assisted detection, analysis and response to security events.

Digital twin: A data-connected digital representation of a physical asset, process or environment.

Decentralised identity: An identity approach in which subjects can control and present verifiable credentials without relying on one central identity store.

Smart contract: Program code that automatically executes rules on a distributed platform.

Trust anchor: A foundational component or authority on which other security decisions depend.

Zero trust: A security approach that continuously evaluates access rather than trusting users or devices solely because of network location.

Graceful degradation: The ability of a system to continue safely at reduced functionality during failure or uncertainty.

Model poisoning: The manipulation of training data or model-development processes to produce harmful behaviour.

Supply-chain assurance: The assessment and control of risks arising from technology providers, components and service dependencies.

Review Questions

Short Answer Questions

1. What security gap does confidential computing address?

2. Why is cryptographic inventory an essential first step in post-quantum preparation?

3. Define network slicing and identify one associated security concern.

4. What is model drift in an edge-AI system?

5. Why should a digital twin report data freshness and confidence?

Long Answer Questions

1. Explain the operating process of a trusted execution environment and critically evaluate its limitations.

2. Discuss how 5G and 6G alter enterprise security responsibilities and supply-chain risks.

3. Analyse the benefits and risks of autonomous security in a critical business environment.

4. Explain the role of cryptographic agility in a long-term post-quantum migration programme.

5. Describe the six-stage framework for developing an adaptable future security roadmap.

Analytical / Application Questions

1. A hospital wants to use confidential computing for collaborative analytics and edge AI for bedside monitoring. Identify the principal trust assumptions, risks and governance controls it should assess before deployment.

2. An industrial company proposes automatic shutdown whenever an edge-AI model detects an anomaly. Design an action-governance model that balances response speed, safety, human oversight and graceful degradation.

3. An enterprise is considering a blockchain for supplier traceability, a digital twin for operational resilience and a PQC migration. Using the chapter’s principles, prioritise these initiatives and justify the sequence.

22. Multiple Choice Questions with Answers

Question 1

Which problem is confidential computing primarily designed to address?

A. Protecting data while it is being processed

B. Replacing all identity-management systems

C. Preventing every form of denial-of-service attack

D. Guaranteeing that application logic is correct

Correct Answer: A. Protecting data while it is being processed

Question 2

What is the most appropriate first activity in a post-quantum migration programme?

A. Immediately remove all existing public-key algorithms

B. Inventory where cryptographic algorithms and keys are used

C. Purchase a quantum computer for testing

D. Move all data to a public blockchain

Correct Answer: B. Inventory where cryptographic algorithms and keys are used

Question 3

Which feature of 5G most directly creates a need for strong isolation between logical services sharing infrastructure?

A. Network slicing

B. Password expiration

C. Offline backup

D. Static routing

Correct Answer: A. Network slicing

Question 4

An edge-AI device begins making inaccurate decisions because factory conditions have changed. Which issue is most directly indicated?

A. Remote attestation

B. Model drift

C. Cryptographic agility

D. Ledger consensus

Correct Answer: B. Model drift

Question 5

Which autonomous security action is generally most suitable for full automation?

A. Permanently deleting evidence from a critical investigation

B. Changing a safety-critical control without review

C. Adding a temporary, reversible detection rule

D. Terminating a major supplier contract

Correct Answer: C. Adding a temporary, reversible detection rule

Question 6

Why can a digital twin create false confidence if its sensor data are compromised?

A. It may represent an inaccurate condition of the physical environment

B. It automatically eliminates all physical safety controls

C. It prevents all forms of network segmentation

D. It makes encryption mathematically impossible

Correct Answer: A. It may represent an inaccurate condition of the physical environment

Question 7

Which statement best describes a limitation of distributed ledgers?

A. They guarantee that every recorded input is factually accurate

B. They eliminate the need for governance and dispute resolution

C. They can preserve records without proving that the original data were correct

D. They cannot support any form of auditability

Correct Answer: C. They can preserve records without proving that the original data were correct

Question 8

What principle should guide the introduction of high-impact autonomous security actions?

A. Use the highest possible automation regardless of consequences

B. Match autonomy to confidence, reversibility and potential harm

C. Avoid recording the reasoning behind automated decisions

D. Give every automated system unrestricted administrative access

Correct Answer: B. Match autonomy to confidence, reversibility and potential harm

Question 9

Which activity belongs to the strategic context stage of a future security roadmap?

A. Mapping critical services, stakeholders and regulatory obligations

B. Deploying an untested model directly into production

C. Disabling all supplier access permanently

D. Replacing every device without risk classification

Correct Answer: A. Mapping critical services, stakeholders and regulatory obligations

Question 10

Northstar rejected a blockchain project because a governed supplier database met its traceability need more simply. What strategic principle does this demonstrate?

A. Every emerging technology should be adopted before competitors do

B. Technology choices should follow clearly defined risks and business outcomes

C. Decentralised systems can never provide useful security

D. Traceability is unrelated to security strategy

Correct Answer: B. Technology choices should follow clearly defined risks and business outcomes