Request an introduction

Every family will have an engineer

The way it has a private banker, a lawyer and a doctor. As AI gets more capable, security is joining that list. This is private cybersecurity, and what it looks like done properly.

HardlineSydney & Melbourne · 8 min read
Darling Harbour, Sydney
Darling Harbour, Sydney.Behind every window, someone's whole life runs through a phone, an inbox and a password.

For most of the internet's history, breaking into one particular person's life was slow, skilled work. Someone had to study you, find a weak point and build a way in. It took weeks of a capable person's time, so very few people were ever worth the trouble. Being hard to reach was a kind of protection.

That protection is going. In the last two years, AI stopped being something attackers talked about and became something they use. The companies that build these models now publish reports on how criminals are using them.

A state-backed group used an AI agent to run most of an espionage campaign

About 30 organisations were targeted. Anthropic estimates the AI did 80–90% of the work, with people stepping in at only a handful of points.

thehackernews.com
One criminal used an AI coding agent to steal from and extort 17 organisations

Healthcare providers, emergency services and government bodies, within a month. Ransom demands ran up to US$500,000.

malwarebytes.com
Google's AI found 20 security flaws in widely used software, on its own

Software that runs inside countless phones, computers and websites. A person checked each report; the AI found and reproduced them.

techcrunch.com
A finance worker joined a video call where everyone else was a deepfake

The "chief financial officer" and colleagues at Arup were all fake. He made 15 transfers totalling about US$25.6 million.

scmp.com
Two years of AI and security.Each of these was a first when it happened. None of them would make the news now.

Those attacks were aimed at companies, and they took some planning. The same tools are now cheap enough to point at a family, and this is the beginning, not the peak. Each new model is noticeably more capable than the last, and the cost of an attempt keeps falling toward nothing. When trying costs nothing, everyone with something worth taking gets tried, all the time.

You're easier to find than you think

Every attack starts with research, and software does that research in seconds. Here's what this page worked out about you in the time since you arrived, without asking you anything.

Visitor ···· ···· ····Since you arrived 0:00
Where you are
Your network
Your device
Your screen
Your time
Your languages
Your software
Your extensions
Your battery
How you got here
Your fingerprint
What this page knows about youWorked out in a few milliseconds and shown only to you. We don't store any of it. A scammer's page would do the same quietly, and keep it.Your IP address is looked up with ipinfo.io, as most websites do without telling you.

The last line is the one that matters. It isn't a cookie, so clearing your history doesn't remove it. Any site that runs the same kind of script sees the same ID, which is how your visits get stitched together into one picture of you.

news-site.com.au/education"Is private school worth $45,000 a year?"ID
online-store.com/gifts/girls-12Birthday gift for a 12-year-oldID
travel-booking.com/flightsSydney → Queenstown, 6–14 July, 4 adultsID
Profile
  • HouseholdChildren at private school
  • FamilyDaughter turning 12
  • Away6–14 July, whole family
  • HouseLikely empty those dates
No cookies. No login. Matched by device.
Followed, without cookies.Your ID from above, as three unrelated sites would see it. Matched on that ID alone, three ordinary visits become a profile worth acting on. Illustrative.

Add a name, which is one search away, and that's the start of a file on you.

Your whole life is the target

Companies spent the last decade building security teams: someone who knows every system, watches it overnight and takes the call when something breaks. Families got none of that, even though the people with the most to lose run their money, their reputation and their children's lives from a handful of phones and inboxes.

An attacker doesn't see a person. They see every place that person can be reached, and they only need one.

VENDORS AND INTEGRATIONSACCOUNTSHOUSEHOLDYour familyPhishing emailSIM swapFake DM, sextortionPassword reuseCamera takeoverDeepfake voice callFake parcel textInvoice fraudStolen API keysMalicious updateYYouPPartnerDDaughterSSonAAssistantHmHouse managerAccountantAI toolsSchool appProperty managerIT contractorTravel agentBrowser extensionsBuilding intercom
Household
YYouPPartnerDDaughterSSonAAssistantHmHouse manager
Accounts
Vendors and integrations
AccountantAI toolsSchool appProperty managerIT contractorTravel agentBrowser extensionsBuilding intercom
Where attacks land
  • Phishing email→Gmail
  • SIM swap→WhatsApp
  • Fake DM, sextortion→Snapchat
  • Password reuse→Bank
  • Camera takeover→Ring
  • Deepfake voice call→Assistant
  • Fake parcel text→Son
  • Invoice fraud→Accountant
  • Stolen API keys→AI tools
  • Malicious update→Browser extensions
A family, as an attacker sees it.Every person, every account, and every company with access to either. The red arrows are common attacks and where they land. Hover or tap anything to see how it's used against you.

Each of these is defended, if at all, by a different app, a different company and a different person, and nobody is looking at the whole picture. That gap is where people get hurt: the email that looks like your accountant, the call that sounds like your son, the router at the beach house that nobody has touched since it was installed.

And it isn't only your own accounts. Every company you've given access to is another way in: the accountant's software, the school app, the property manager, the AI tool someone connected to your email. People forget these exist. Attackers don't.

OpenCode breach drains users' AI keys

Attackers reached a legacy system and took stored OpenAI, Anthropic and Google keys for about 1.4% of its users. Some developers reported hundreds of dollars in charges overnight.

dealroom.co
This week.The users did nothing wrong. A company they'd trusted with their keys was breached.

It's already happening here

Australians report a cybercrime to the government every six minutes, more than 84,700 times last financial year. Since AI tools went mainstream, the average loss for an individual has risen every year.

$26,200
$30,700+17%
$33,000+8%
2022–232023–242024–25
Average loss per reported cybercrimeChatGPT weekly users, launch to 500 million
Losses rose as AI took offAverage loss per reported cybercrime for Australian individuals, by financial year, against ChatGPT's weekly users over the same period. One doesn't prove the other, but the tools and the losses grew together. ChatGPT passed 800 million weekly users later in 2025.Australian Signals Directorate, Annual Cyber Threat Reports 2023–24 and 2024–25; the 2022–23 figure is implied by the 17% rise ASD reported. ChatGPT figures from OpenAI announcements, via TechCrunch.

Those are only the crimes people chose to report, and the families with the most to protect are often the least willing to. Meanwhile, the raw material for identity theft and impersonation is already out there. In the last few years, the personal details of most Australians have been taken from companies they trusted.

  1. OptusNames, dates of birth, addresses and ID numbers of current and former customers.Up to 9.8 million people
  2. MedibankPersonal and health claims data, later published online.9.7 million customers
  3. Latitude FinancialCustomer records, including driver's licence and passport numbers.About 14 million records
  4. MMediSecureNames, addresses, dates of birth and Medicare details from prescriptions.12.9 million Australians
Four breaches, most of the countryYour phone number, address and date of birth are almost certainly in at least one of these, and copies circulate for years.Company statements, as reported. Latitude, MediSecure.

Put that together with what's already public about you, and software has everything it needs to write to you as your bank, or call your assistant as you.

It isn't really a question of whether someone will try. It's when, and whether anyone is there when they do.

How private cybersecurity works

The usual answer is more software: an antivirus licence, a password manager, another dashboard. They help, but each one guards a single door, and none of them picks up the phone. Hardline works the other way around. Your family gets a senior engineer of its own, the way a family office gives you one person who knows your money.

Wk 0Wk 1Wk 2Wk 3Wk 4Wk 5Wk 6· · ·OngoingYear 1
IntroductionA private call, under NDA
Scope agreed
DiscoveryAccounts, devices, homes and people
Two weeks
ReportRanked by real-world risk
Delivered, yours to keep
FixCritical findings first
Critical in 72 hours, the rest by week six
MembershipConcierge and monitoring
Your engineer, any hour, every day
RecoveryOnly if something gets through
Same-day response, included
Yearly reviewEverything checked again
Every year
How an engagement runs.Two weeks to see everything, about six to fix it, then a membership that runs for as long as you need it.

Discovery ends with a written report, ranked by how each finding could actually be used against you. We start closing the worst of it on the first day.

A page from the first report.Every finding ranked and closed, one by one. You keep the report whether or not you continue. A sample, not a real household.

After that, you're a member. This is everything your engineer looks after, for every person and every home.

What's includedHardlineSecurity appAntivirus, identityIT supportCall-out, managed
Concierge
A senior engineer of your ownThe same person every time, who knows your household.
Any hour, by phone or messageAnswered in minutes, nights and weekends included.
"Is this real?" checksSend any email, text, call or invoice before you act on it.
New devices, set up properlyPhones and laptops configured, old ones wiped and retired.
Travel preparationDevices readied before you leave; someone on call while away.
Monitoring
Account sign-insNew logins, password resets and recovery changes on accounts that matter.
Breaches and the dark webYour family's emails, numbers and passwords, watched for in new leaks.
ImpersonationFake profiles, lookalike sites and scam pages using your name.
Every home, overnightNew devices and unusual traffic on each residence's network.
Vendors and integrationsWhich companies and apps hold access to your accounts, and what they can do.
Recovery
Contained within the hourPasswords, sessions, cards and devices locked down.
Hijacked accounts recoveredEmail, social, Apple and Google.
Banks, police, insurers, platformsEvery call and form, handled for you.
Never billed by the hourRecovery is part of the membership.
Money
Call-back rules with your bankNew or large payments confirmed by phone, every time.
Approval rules for staffSo no single email can move money.
Payment alertsUnusual transfers flagged to you and your engineer.
Accounts and passwords
Family password managerWith shared vaults for the household.
Two-step login everywhereApp or hardware key, not text messages.
Leaked passwords replacedStarting with email and banking.
Recovery locked downSo nobody can reset their way into your accounts.
Privacy
Data-broker removalYour family taken off people-search sites, and kept off.
Public records tidiedProperty and company records, where the law allows.
Children's footprintSchool, sport and social posts reviewed with you.
Deepfake protectionFamily code words and verification routines for voice and video.
Homes and devices
Networks rebuiltRouters and Wi-Fi replaced or reconfigured, then kept updated.
Cameras, locks and smart devicesMoved off the family's network, passwords changed, updated.
Every device kept patchedPhones, tablets and laptops for the whole household.
Family and staff
BriefingsShort, practical sessions for partners, children and staff.
Staff joining and leavingAccess given on the first day and removed on the last.
Yearly reviewEverything checked again, with a fresh report.
IncludedSometimes, or in partNot usually
The membership, in full.For every person and every home in the household. Comparisons are typical of consumer security apps and IT support providers, not any particular company.

What it's like

Most of it you never see. When something does land in front of you, like an email late at night that looks exactly like it came from your private bank, you don't have to work it out alone.

  1. 1A lookalike sender
  2. 2A deadline, so you rush
  3. 3A link to a fake login page
A typical lure.Written by software, personalised from what's public, and sent at night. The bank and sender are fictional.

You forward it and ask one question. And more often, you won't need to: the first you hear of a problem is your engineer telling you it's already been dealt with.

What it's like.Left: your engineer gets there first and tells you afterwards. Right: you forward something suspicious and have an answer in four minutes, from someone who already knows your accounts, your staff and your homes. Examples, not real clients.

Your engineer is close enough to be at your door within the day.

Sydney
Sydney
Melbourne
Melbourne
Where we workOn site within the day in Sydney and Melbourne. Brisbane by arrangement.

Pricing

It's priced against what you could lose, not against software, and it begins with discovery, so you see everything we find before committing to anything.

What a year looks like.One line for the membership. Recovery, extra homes and family members are included.
Request an introductionYour company can pay on your behalf. NDA on request.

Introductions

We don't advertise, and we don't publish who we work for. Most families come to us through someone they trust. If that isn't you, start here: a short, private call, under NDA if you'd like, before you tell us anything about your household.

Hardline

Private cybersecurity. Like private banking, for your digital life. Sydney and Melbourne.

Call [YOUR NUMBER][YOUR EMAIL]