Red Asia Group · Leadership report · November 2026
Cyber risk in the age of AI agents
A 2026 cyber security briefing for business leaders in Indonesia and Southeast Asia: how attacks now work, what they cost, what AI changes, and what to do in the next 90 days.
- 8 chapters
- About 60 minutes in full
- Evidence to 17 Nov 2026
Start here
What cyber risk means for a business
Cyber risk is the chance that an attack or failure in your digital systems stops operations, exposes data or loses money, and what that would cost. It is a business risk, measured like any other: how likely, and how bad.
Go deeperWhy this is no longer an IT budget line
For most of the last decade, boards treated cyber security as an IT budget line. The cases in this report show why that no longer works. When a British carmaker's systems went down in September 2025, its factories stood still for about five weeks, the government guaranteed £1.5bn of lending to keep its suppliers alive, and the wider UK economy lost an estimated £1.9bn. 10,11,12
Two things changed at once. Criminals now use AI to run parts of an attack that used to need a team of people, and companies are giving AI agents real authority to read email, update customer records and move money. An agent with that access can be talked into things by a document it reads.
This report is written for the people who answer for the consequences: owners, CEOs, CFOs and directors. Where a part is technical, it says so and tells you who in your company should read it.
The thread through this report
10 questions your board will ask
Every part of this report answers some of them. By the end you can answer all ten, and the self-check at the end asks them back.
What can each of our AI agents actually do, and can we stop it?
Answered in: AI agents
Who owns each supplier and app connection into our systems?
Answered in: How attacks work
Can one person both request and approve a payment?
Answered in: Cost
Could an email or document trigger an action in our systems?
Answered in: AI agents
Who must we tell after a breach, and how fast?
Answered in: Indonesia
Have we proven we can restore our critical systems?
Answered in: How attacks work, Indonesia
Would our finance team stop a convincing fake request from the CEO?
Answered in: Cost
What happens if a critical supplier goes down?
Answered in: Timeline
Which risks have we accepted, and who owns them?
Answered in: Action plan
Your route through the report
Three stages, about an hour in full. Each page shows only the essentials first; open “Go deeper” where you want the detail. The bar at the top of every page shows where you are and which questions you have covered.
1. Understand the threat
2. See your exposure
Try the tools
Six interactive tools from inside the report, ready to use here. Each has its own link you can send to a colleague.
2 min
The fraud drill
Three moments from a real fraud pattern. Would your finance team pay?
5 min
The board self-check
Ten questions, four honest answers, and where to start.
3 min
The loss calculator
What a few days of downtime would cost your business, in Rupiah.
1 min
Map an AI agent's reach
Tick what an assistant may do and watch what one misuse could touch.
1 min
See prompt injection
A hidden line in a document makes an assistant leak your pricing.
2 min
The 72-hour clock
What should happen, hour by hour, after a personal data breach in Indonesia.
Six numbers, read correctly
Each figure comes from a different study of a different population. Tap any number to see exactly what it measures before you quote it.
Tap “What this measures” under any figure
- 1
31%
of breaches with a known entry point began with an exploited software flaw, now the top route in.
- 1,2
48%
of breaches involved a supplier or other third party, up from 30% a year earlier.
- 7
29 min
average time for criminals to spread beyond the first machine they break into.
- 4,54
US$4.12m
average cost of a data breach for an organization in Southeast Asia, a record high.
- 5
US$3.05bn
lost to fake payment instructions sent by email in 2025, in complaints to the FBI alone.
- 43
72 hrs
to notify people and the authority after a personal data breach in Indonesia.
Go deeperHow breaches changed in one year
Exploited software flaw as the way in
20% → 31% Now the most common known entry route, ahead of stolen passwords.
Ransomware present
44% → 48% Almost half of all breaches.
Supplier or third party involved
30% → 48% The fastest-growing measure in the study.
A person involved (a click, a call, a mistake)
60% → 62% Still the majority.
Shares of breaches in each year's dataset; categories overlap, so they do not add up to 100%. The first row counts only breaches whose entry route was known. Sources: 1,2
Six findings that should shape your decisions
Old weaknesses still open the door
Exploited software flaws, stolen passwords and persuasive phone calls account for most of the breaches in the 2026 data. 1 AI has not replaced these routes; it makes them cheaper to use at scale.
The help desk is now a front door
Attackers phone IT support pretending to be staff, then ask for a password reset or a new sign-in device. US and UK agencies describe this as a standard technique of the group behind the 2025 UK retail attacks. 14
Your own AI agents are a new attack surface
An assistant that reads untrusted email and can also send data out can be tricked into leaking it. A real flaw of this kind was found and fixed in a major office assistant in 2025. 27
Five decisions for your next board meeting
None of these requires choosing a security product. Each has an owner and a piece of evidence you can ask to see.
| Decision | Who owns it | Evidence to ask for |
|---|---|---|
| Name the services the company cannot run without, and how long each can be down | CEO and COO | A ranked list with a tolerable outage time for each, signed off by the business owners |
| Set limits on what AI agents may do | CIO with business owners | A register of every agent, what it can access, and proof that its riskiest actions need human approval |
| Protect sign-in and supplier access | CISO or Head of IT | Coverage of strong sign-in on admin and remote access, and a test of cutting off one supplier's access |
| Rehearse an outage and a fake payment request | COO and CFO | Results of a restore test and a payment-fraud exercise, including what failed |
| Fund the gaps that matter most | CEO and CFO | A costed plan tied to the services named in the first decision |
Questions leaders ask
What is 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. It is managed like any business risk: by likelihood and impact.
Why is cyber security a CEO and board issue, not just IT?
Because the consequences are business ones: halted production, lost sales, fines, fraud and lost customers. In 2025, one attack stopped a carmaker's factories for about five weeks and needed a £1.5bn government-backed loan guarantee to protect its suppliers. 10,11
Who is this report for?
CEOs, owners, CFOs and directors in Indonesia and Southeast Asia who make decisions about cyber risk without a technical background. Technical parts are marked so you can forward them to your CIO, CTO or Head of IT.
Where should a leader start?
With three questions: what would stop the business, who is in charge when it happens, and how fast can we recover. The 90-day plan in this report turns those into tasks with owners.
How current is this report?
It uses public evidence up to 17 November 2026 and was published on 17 November 2026. It will be updated when major new evidence changes a finding.







