Chapter 7 · Glossary and sources
Cyber security, in plain words
The terms your security team will use, explained by what they mean for the business. Then how this report was put together, and a link to every original source.
- 4 min read
- 38 terms
- 60 sources
Glossary
- AI agent
Software that uses an AI model to take actions on its own, such as sending messages, changing records or moving money.
Why it matters: It needs its own identity, limits and owner, like an employee with system access.
- Agentic workflow
A business process in which an AI system chooses or carries out some of the steps.
Why it matters: Decide which steps it may take alone and which need a person.
- Attack surface
Everything an attacker could try to use to get in: systems facing the internet, accounts, suppliers, apps and people.
Why it matters: Every new system, app or agent adds to it.
- Backup (offline or immutable)
A copy of data that attackers cannot reach or change, even with admin rights.
Why it matters: Ransomware deletes ordinary backups first. Only tested, protected copies save you.
- Blast radius
How much damage a single compromised account, system or agent could cause.
Why it matters: Smaller access means a smaller blast radius.
- Breakout time
How long an attacker takes to move from the first machine they compromise to others. The 2025 average was 29 minutes. 7
Why it matters: It sets how fast your response must be.
- Business email compromise (BEC)
Fraud in which criminals pose as an executive or supplier, usually by email, to redirect a payment.
Why it matters: It costs billions a year and needs no hacking at all. 5
- CISO
Chief Information Security Officer: the executive responsible for security. In smaller firms, often the Head of IT.
Why it matters: Someone must own security with real authority and a line to the CEO.
- Credential
Anything that proves identity to a system: a password, a code, a key or a token.
Why it matters: Stolen credentials are among the most common ways in.
- Cyber attack
A deliberate attempt to break into, disrupt or misuse an organization's systems or data.
Why it matters: Most aim at money, information, or access to someone else.
- Cyber risk
The chance that an attack or failure in your digital systems stops operations, exposes data or loses money, and what that would cost.
Why it matters: It is a business risk, measured by likelihood and impact.
- Data breach
An incident in which data is accessed, taken or exposed without authorization.
Why it matters: If personal data is involved in Indonesia, the 3 × 24-hour notice duty can apply. 43
- Deepfake
Fake audio, images or video generated by AI to look and sound like a real person.
Why it matters: A familiar face on a call is no longer proof of identity. 24
- Encryption
Scrambling data so only someone with the right key can read it.
Why it matters: It protects your data. Ransomware turns the same technique against you.
- Endpoint
Any device that connects to your network: laptops, phones, servers.
Why it matters: Most attacks pass through one.
- Exfiltration
Copying data out of a company without permission.
Why it matters: Attackers steal before they encrypt, so backups alone do not stop extortion.
- Incident response plan
A written, rehearsed plan for who does what during a cyber incident.
Why it matters: It saves the first critical hours.
- Lateral movement
An attacker moving from the first compromised system to others inside your network.
Why it matters: Stopping it early keeps an incident small.
- Least privilege
Giving each person, system or agent only the access its job needs.
Why it matters: It caps the blast radius of any one compromise.
- Malware
Malicious software of any kind, including viruses, password stealers and ransomware.
Why it matters: Password-stealing malware on a personal laptop can open company accounts. 19
- Managed security service (MSSP / MDR)
An outside provider that watches your systems and responds to threats, often around the clock.
Why it matters: A practical way to cover nights and weekends.
- MCP (Model Context Protocol)
A common way to connect AI models to tools and data sources.
Why it matters: It is a connector, not a guarantee of safety. Review each connection's permissions.
- Multi-factor authentication (MFA)
Signing in with a second proof, like an app prompt or a security key, as well as a password.
Why it matters: It stops most password-theft attacks. Missing MFA on one portal cost one company billions. 17,18
- OAuth token
A digital pass that lets one app act on your behalf inside another, without your password.
Why it matters: Stolen tokens let attackers in through connected apps. 26
- Patch
A software update that fixes a security flaw.
Why it matters: Exploited flaws are now the top known way in. 1
- Penetration test
An authorized, simulated attack to find weaknesses before criminals do.
Why it matters: It shows what is exploitable, not just what is theoretically weak.
- Phishing
A fake message designed to trick someone into clicking, signing in, sharing data or paying.
Why it matters: It remains the most common start of an attack.
- Phishing-resistant MFA
Sign-in methods, such as passkeys and security keys, that a fake login page cannot capture. 58
Why it matters: Ordinary codes can be phished; these cannot.
- Prompt injection
Hiding instructions in content an AI reads, such as an email or document, so that it does something its owner did not intend.
Why it matters: Reading a document should never give it authority. 39
- Ransomware
Malicious software that locks or steals a company's data, followed by a demand for payment.
Why it matters: It was present in 48% of breaches in the 2026 data. 1
- RTO and RPO
Recovery time objective: how quickly a service must be back. Recovery point objective: how much recent data you can afford to lose.
Why it matters: The business sets these; IT designs recovery to meet them.
- Security operations center (SOC)
A team that monitors systems and responds to attacks, in-house or outsourced.
Why it matters: It is how alerts become action at 2 a.m.
- Shadow AI
AI tools staff use without approval or oversight.
Why it matters: Data put into them can leave the company. Make it easy to declare tools rather than hide them.
- Supply-chain attack
An attack that reaches you through a supplier, software vendor or connected service.
Why it matters: Third parties were involved in 48% of breaches in the 2026 data. 1
- Vulnerability (zero-day, n-day)
A weakness in software. A zero-day has no fix yet; an n-day is already public and fixable.
Why it matters: Most damage comes from n-days left unpatched.
- VPN and remote access
Ways for staff and suppliers to reach company systems from outside.
Why it matters: They are frequent targets; protect them with strong sign-in and patching. 15
- Zero trust
An approach where no user, device or agent is trusted by default; every access is checked every time.
Why it matters: It limits what an attacker can do after getting one foothold.
How this report was made
Sources
Public evidence up to 17 November 2026: company filings and statements, government agencies and regulators, original research and security-provider investigations. Cases were chosen because each teaches a leader something specific, not as a statistical sample.
Figures
Each figure keeps its source, date and population. Figures from different studies are never added together, and lab results are never presented as the odds of a real breach.
Attribution
Where a company or agency attributes an attack to a group or to AI, this report says who made the assessment. Where a cause is only reported in the press, it says so.
Scenarios
The loss calculator, the fraud drill and the prompt injection demonstration are illustrations, labelled as such. They are not forecasts or reported incidents.
Law
Legal and regulatory sections summarize obligations for business leaders as of November 2026. They are not legal advice.
Updates
Public reports change as investigations continue. This report will be updated when new evidence changes a finding.
Sources
Every source cited in the report, numbered as in the text. All links checked on 17 November 2026.
- [1]
Verizon · 2026 Data Breach Investigations Report, executive summary
May 2026. More than 22,000 confirmed breaches in 145 countries. The 31% exploitation figure is a share of breaches where the initial access route was known.
Open the original - [2]
Verizon · 2025 Data Breach Investigations Report, executive summary
April 2025. Used for the year-on-year comparison only.
Open the original - [3]
IBM · Study: one in four malicious breaches are AI-enabled, costing companies $6 million on average
29 July 2026. 602 organizations breached between March 2025 and February 2026.
Open the original - [4]
IBM and Ponemon Institute · Cost of a Data Breach Report 2026
July 2026. Global average US$4.99m. The ASEAN figure (US$4.12m) comes from a sample of 26 organizations.
Open the original - [5]
FBI Internet Crime Complaint Center · 2025 Internet Crime Report
April 2026. Complaints made to the FBI, including consumer fraud. Business email compromise: US$3.05bn from 24,768 complaints.
Open the original - [6]
World Economic Forum · Global Cybersecurity Outlook 2026
January 2026. 804 qualified participants from 92 countries. Self-reported processes, not audited results.
Open the original - [7]
CrowdStrike · 2026 Global Threat Report
24 February 2026. Breakout time is how long criminals take to move from the first machine to others, observed in CrowdStrike's own cases.
Open the original - [8]
CrowdStrike · 2025 Global Threat Report findings
27 February 2025. 48-minute average breakout in 2024.
Open the original - [9]
Marks & Spencer · Full-year results, 52 weeks ended 28 March 2026
20 May 2026. Incident costs of £131.3m and insurance proceeds of £100m.
Open the original - [10]
Jaguar Land Rover · JLR performance impacted in challenging quarter
November 2025. £196m of cyber-related costs in the July to September 2025 quarter.
Open the original - [11]
UK Government · Government backs Jaguar Land Rover with £1.5 billion loan guarantee
28 September 2025. Loan guarantee to stabilize a supply chain employing about 120,000 people.
Open the original - [12]
Cyber Monitoring Centre · Statement on the Jaguar Land Rover cyber incident
22 October 2025. Modeled UK economic loss of about £1.9bn; more than 5,000 UK organizations affected.
Open the original - [13]
Co-operative Group · 2025 trading update
2026. Estimated £285m revenue impact and £107m profit impact from the 2025 attack.
Open the original - [14]
CISA, FBI and partners · Scattered Spider joint advisory (AA23-320A)
Revised 29 July 2025. Describes criminals calling IT help desks to have passwords reset and sign-in devices moved.
Open the original - [15]
Asahi Group Holdings · Update on the system disruption caused by a cyberattack
18 February 2026. Entry through network equipment at a group site; a password weakness gave administrator rights.
Open the original - [16]
RTX · Form 8-K: Collins Aerospace MUSE incident
19 September 2025. Ransomware on systems that run shared airport check-in and baggage software.
Open the original - [17]
UnitedHealth Group · Testimony of Andrew Witty to the US Senate Finance Committee
1 May 2024. A remote-access portal without multi-factor sign-in; a US$22m ransom.
Open the original - [18]
UnitedHealth Group · 2024 results
16 January 2025. About US$3.09bn of 2024 cyberattack effects.
Open the original - [19]
Google Mandiant · UNC5537 targets Snowflake customer instances
June 2024. About 165 organizations notified; stolen passwords, no multi-factor sign-in.
Open the original - [20]
MGM Resorts · Form 8-K
5 October 2023. About US$100m impact on adjusted property EBITDAR.
Open the original - [21]
Qantas · Update on July cyber incident
2025. 5.7m customer records taken from a third-party platform used by a contact center.
Open the original - [22]
Delta Air Lines · Form 8-K
August 2024. About 7,000 cancellations over five days and at least US$500m in losses after the CrowdStrike update failure.
Open the original - [23]
Microsoft · Helping our customers through the CrowdStrike outage
20 July 2024. 8.5m Windows devices affected.
Open the original - [24]
Hong Kong Government · LCQ9: Combating frauds involving deepfake
26 June 2024. A pre-recorded fake video conference and about HK$200m in losses.
Open the original - [25]
CNN Business · Arup revealed as victim of US$25 million deepfake scam
16 May 2024. Arup confirmed fake voices and images were used.
Open the original - [26]
Google Threat Intelligence Group · Widespread data theft targets Salesforce instances via Salesloft Drift
27 August 2025. Stolen app tokens, active 8 to 18 August 2025.
Open the original - [27]
Microsoft Security Response Center · CVE-2025-32711 (EchoLeak)
June 2025. A flaw in Microsoft 365 Copilot found by Aim Labs and fixed by Microsoft; no customer data was accessed.
Open the original - [28]
Anthropic · Disrupting the first reported AI-orchestrated cyber espionage campaign
13 November 2025. About 30 targets; AI did 80 to 90% of the tactical work, in Anthropic's assessment.
Open the original - [29]
Anthropic · Detecting and countering misuse of AI: September 2026
September 2026. Activity disrupted from December 2025 to August 2026.
Open the original - [30]
Anthropic · Detecting and countering misuse of AI: August 2025
27 August 2025. Data extortion against at least 17 organizations using an AI coding agent.
Open the original - [31]
OpenAI · The Hugging Face incident and the road ahead
26 August 2026. Evaluation agents escaped isolation and reached Hugging Face systems in July 2026.
Open the original - [32]
METR and Redwood Research · Independent investigation of the OpenAI / Hugging Face incident
26 August 2026. About 1,200 agents on an unsanctioned message board; about 700 took part in the attack.
Open the original - [33]
Google Threat Intelligence Group · GTIG AI Threat Tracker: Agentic threat activity crosses a significant threshold
17 November 2026. Attack frameworks built on AI agents observed in real intrusions.
Open the original - [34]
Fang, Bindu, Gupta and Kang · LLM agents can autonomously exploit one-day vulnerabilities
April 2024. A test of 15 known flaws; a lab result, not a breach rate.
Open the original - [35]
Zhu and coauthors · CVE-Bench: AI agents and real-world web application vulnerabilities
March 2025. Up to 13% of vulnerabilities exploited in its test setup.
Open the original - [36]
NIST · Design tenets for agent identity management systems
27 October 2026. Ten proposed principles; not a standard.
Open the original - [37]
CISA, NSA and partners · Careful adoption of agentic AI services
1 May 2026. Joint guidance from US, Australian, Canadian, New Zealand and UK agencies.
Open the original - [38]
NSA, CISA, FBI and partners · AI data security: best practices for securing data used to train and operate AI systems
22 May 2025. Lifecycle guidance for AI data.
Open the original - [39]
NIST · Adversarial machine learning: a taxonomy (AI 100-2 E2025)
March 2025. Prompt injection, poisoning, evasion and privacy attacks.
Open the original - [40]
UK National Cyber Security Centre · Impact of AI on cyber threat from now to 2027
7 May 2025. An intelligence assessment, not a count of attacks.
Open the original - [41]
Cyber Security Agency of Singapore · Advisory on recent uptick in reported cyber incidents (AD-2026-017)
13 November 2026. Monthly reported incidents since September at three times the January to August average.
Open the original - [42]
Cyber Security Agency of Singapore · Singapore Cyber Landscape 2025/2026
30 June 2026. 165 ransomware cases reported in 2025.
Open the original - [43]
Republic of Indonesia · Law No. 27 of 2022 on Personal Data Protection (UU PDP)
17 October 2022. Article 46 (3 × 24-hour notice), Article 34 (impact assessment), Article 56 (transfers abroad), Article 57 (administrative sanctions).
Open the original - [44]
Katadata · Komdigi: presidential regulation on the PDP supervisory agency at final stage
5 November 2026. The agency has yet to be formally established.
Open the original - [45]
Katadata · PDP implementing regulation takes effect January 2027
2026. PP 33/2026, set 16 July 2026, in force 16 January 2027.
Open the original - [46]
Otoritas Jasa Keuangan (OJK) · SEOJK 29/SEOJK.03/2022 on cyber resilience and security for commercial banks
December 2022. Initial notice to OJK within 24 hours; full report within 5 working days.
Open the original - [47]
Bank Indonesia · PBI 2/2024 on information system security and cyber resilience
18 April 2024. Applies to payment-system operators and other BI-supervised firms.
Open the original - [48]
Republic of Indonesia · Presidential Regulation 47/2023 on the national cyber security strategy and cyber crisis management
20 July 2023. Electronic system operators must act on BSSN early warnings.
Open the original - [49]
DPR RI · Commission I speeds up deliberation of the cyber security and resilience bill (RUU KKS)
5 July 2026. A priority bill; not yet passed in November 2026.
Open the original - [50]
Kominfo · Pemulihan PDNS 2: beberapa layanan publik berangsur pulih
26 June 2024. 282 tenants of the temporary national data center affected.
Open the original - [51]
Republika · BSSN: PDNS data had no backup
27 June 2024. BSSN told parliament about 98% of PDNS 2 data had no backup.
Open the original - [52]
detikFinance · BSI outage: suspected cyber attack, president director explains
May 2023. Services disrupted from 8 May 2023; mobile banking restored 11 May.
Open the original - [53]
Bloomberg Technoz · BSSN: 330.5 million traffic anomalies in 2024
January 2025. Anomalies, not confirmed attacks; BSSN monitors part of national traffic.
Open the original - [54]
CRN Asia · IBM: ASEAN data breach costs hit record US$4.12m
2026. ASEAN sample of 26 organizations; US$3.67m the year before.
Open the original - [55]
Bisnis.com · Cisco research: 89% of Indonesian companies not ready for cyber attacks
26 May 2025. Cisco Cybersecurity Readiness Index 2025: 11% of Indonesian organizations at the mature stage.
Open the original - [56]
Coveware by Veeam · Quarterly ransomware report, Q3 2026
28 October 2026. Median payment US$400,000; nearly 80% of cases resolved without payment.
Open the original - [57]
Sophos · The State of Ransomware 2026
2026. 2,158 organizations hit by ransomware in 17 countries.
Open the original - [58]
CISA · Implementing phishing-resistant MFA
Fact sheet. Sign-in methods that a fake login page cannot capture.
Open the original - [59]
Anderson Economic Group via Motor1 · Dealers lose about US$1 billion after CDK cyberattack
July 2024. An economic estimate of direct dealer losses.
Open the original - [60]
Brussels Airport · 2.3 million passengers in September
October 2025. Flights cancelled 20 to 28 September after the check-in supplier was attacked.
Open the original
Questions leaders ask
What is a SOC?
A security operations center: a team, in-house or outsourced, that watches an organization's systems around the clock and responds to attacks.
What is zero trust?
An approach where no user, device or AI agent is trusted by default, and every access is checked every time, limiting what an attacker can do after one foothold.
What is multi-factor authentication?
Signing in with a second proof, such as an app prompt or security key, as well as a password. Phishing-resistant forms such as passkeys cannot be captured by a fake login page. 58
