Threat actors
Attackers differ by motive, skill and money, and so must the defence
Threat actors range from unskilled attackers to organised crime and nation-states; insiders start inside.
Who is on the other side? Place them on a line from little skill and money to a great deal. At one end, the unskilled attacker, running tools others wrote, for the thrill. Basic hygiene stops them. Hacktivists attack for a cause: they deface sites, leak documents, flood servers. Insiders are different. Skill varies, but they are already inside, with real access. Organised crime is a business. It wants money and picks the cheapest way in. At the far end, nation-states: patient, well funded, after secrets or disruption, and able to stay hidden for months.
Threat actor attributes · Going deeper
Internal or external, resources, sophistication and motive
Exams describe a threat actor by four attributes and ask you to name it, or by its motivation.
Exams describe a threat actor by four attributes and ask you to name it, or by its motivation. Four attributes. Internal or external insiders and contractors start with access; everyone else must gain it. Resources and funding from a laptop and free tools to a state budget and dedicated teams. Level of sophistication from copying public exploits to finding and using unknown (zero-day) flaws. Motivation what they want, which predicts what they will target. Motivations named in the objectives. Two that are easy to forget. Shadow IT staff using systems or cloud services the organisation has not approved. Not malicious, but unmanaged, unpatched and invisible to security. Advanced persistent threat (APT) an actor, usually state-backed, that gains access and stays undetected for a long time to steal data. “Persistent” is the defining word. An insider is not only a malicious employee: a careless one who emails the wrong file is the more common insider incident.
Four attributes
- Internal or external insiders and contractors start with access; everyone else must gain it.
- Resources and funding from a laptop and free tools to a state budget and dedicated teams.
- Level of sophistication from copying public exploits to finding and using unknown (zero-day) flaws.
- Motivation what they want, which predicts what they will target.
Motivations named in the objectives
Two that are easy to forget
- Shadow IT staff using systems or cloud services the organisation has not approved. Not malicious, but unmanaged, unpatched and invisible to security.
- Advanced persistent threat (APT) an actor, usually state-backed, that gains access and stays undetected for a long time to steal data. “Persistent” is the defining word.
Initial access
Most breaches start with a stolen login, an exposed flaw or a lure
Verizon 2025: credential abuse 22%, exploitation 20%, phishing 16% of breaches began this way.
Whoever the attacker is, they need a way in. Verizon studied over twelve thousand breaches in 2025. The most common start: a stolen or guessed login. Twenty-two percent. Close behind, exploiting a software flaw, often in the very VPNs and firewalls that guard the edge of a network. Then phishing: a message that fools a person. Across all breaches, sixty percent involved a person’s action, forty-four percent included ransomware, and thirty percent came through a third party. Three doors account for most intrusions. Close those, and most attackers move on.
Attack surface · Going deeper
Every door an attacker could try, and how to shrink it
The attack surface is every reachable entry point; threat vectors are the paths attackers use through it.
The attack surface is every reachable entry point; threat vectors are the paths attackers use through it. Common threat vectors. Message-based email, SMS (smishing), instant messages, voice calls (vishing), QR codes (quishing). Unpatched or end-of-life software especially internet-facing VPNs, firewalls and file-transfer servers. Open service ports and default credentials a remote desktop port left open, an admin password never changed. Removable media and images a USB stick left in a car park; a malicious file disguised as a picture. Supply chain managed service providers, vendors and software suppliers who hold trust in your network. Wireless and Bluetooth rogue access points, insecure pairing. Shrinking the surface. Turn off what is not needed, close ports, remove default accounts, patch what faces the internet first, and require a second factor everywhere a password is accepted. The median time to fully patch edge-device flaws in Verizon’s data was 32 days; attackers often exploit them within days of disclosure.
Common threat vectors
- Message-based email, SMS (smishing), instant messages, voice calls (vishing), QR codes (quishing).
- Unpatched or end-of-life software especially internet-facing VPNs, firewalls and file-transfer servers.
- Open service ports and default credentials a remote desktop port left open, an admin password never changed.
- Removable media and images a USB stick left in a car park; a malicious file disguised as a picture.
- Supply chain managed service providers, vendors and software suppliers who hold trust in your network.
- Wireless and Bluetooth rogue access points, insecure pairing.
Shrinking the surface
Turn off what is not needed, close ports, remove default accounts, patch what faces the internet first, and require a second factor everywhere a password is accepted. The median time to fully patch edge-device flaws in Verizon’s data was 32 days; attackers often exploit them within days of disclosure.
The Cyber Kill Chain
An intrusion runs through seven steps, and each is a chance to stop it
The Cyber Kill Chain: reconnaissance, weaponisation, delivery, exploitation, installation, command and control, actions on objectives.
Attacks are not one event. They are a sequence, and the Target breach of 2013 shows it well. Reconnaissance: the attackers found Target’s suppliers listed online. Weaponisation: they prepared malware, probably inside an attachment. Delivery: a phishing email to a heating and cooling supplier. Exploitation: they used the supplier’s login to enter Target’s network on the twelfth of November. Installation: card-stealing malware on the tills. Command and control: a channel to the outside, held open for weeks. Actions on objectives: card data for forty million people sent out. Twice, Target’s tools raised alerts. Nobody acted. Every step was a chance to stop it.
The Target breach · Going deeper · 1 of 6
Five weeks inside, and two ignored alerts, drawn to scale
Alerts on 30 November and 2 December went unanswered; the government told Target on 12 December.
The Cyber Kill Chain · Going deeper · 2 of 6
Fifteen tactics name what attackers are trying to do at each moment
MITRE ATT&CK maps real attacks as tactics (the goal) and techniques (the method), each with an ID.
The kill chain shows the outline. ATT&CK fills in the detail. It is a public catalogue, built from reports of real intrusions. Across the top row: reconnaissance, getting resources, getting in, running code, and staying in. The middle row: gaining more privilege, staying hidden, disabling defences, stealing passwords, and looking around. The last row: moving to other machines, gathering data, talking to the outside, taking data out, and finally the impact, such as encrypting everything. Under each tactic sit techniques with IDs. Defenders use them to say exactly what they can and cannot detect.
The Diamond Model · Going deeper · 3 of 6
Adversary, capability, infrastructure, victim: pivot on any corner
The Diamond Model links four features of every intrusion so analysts can pivot from one to the next.
Tactics, techniques and procedures · Going deeper · 4 of 6
TTPs: how defenders use ATT&CK in practice
TTPs describe how an adversary operates. ATT&CK gives each one a shared ID.
TTPs describe how an adversary operates. ATT&CK gives each one a shared ID. Three levels. Tactic the goal: “credential access”. Technique how the goal is reached: “OS credential dumping, T1003”. Many have sub-techniques, such as T1003.001 for dumping LSASS memory. Procedure the specific way one group does it: a named tool with particular commands. Four everyday uses. Detection coverage colour the matrix by which techniques your logs and tools can detect, and find the gaps. Threat-informed defence look up the groups that target your industry and prioritise their techniques. Red and purple teaming plan tests technique by technique and check each was seen. Reporting incident and threat reports cite technique IDs, so findings can be compared and shared. ATT&CK has three matrices: Enterprise (Windows, macOS, Linux, cloud, network, containers), Mobile and ICS. The free ATT&CK Navigator lets you build and share coverage maps.
Three levels
- Tactic the goal: “credential access”.
- Technique how the goal is reached: “OS credential dumping, T1003”. Many have sub-techniques, such as T1003.001 for dumping LSASS memory.
- Procedure the specific way one group does it: a named tool with particular commands.
Four everyday uses
- Detection coverage colour the matrix by which techniques your logs and tools can detect, and find the gaps.
- Threat-informed defence look up the groups that target your industry and prioritise their techniques.
- Red and purple teaming plan tests technique by technique and check each was seen.
- Reporting incident and threat reports cite technique IDs, so findings can be compared and shared.
The Cyber Kill Chain · Going deeper · 5 of 6
Block what attackers do, not the files they use, and it costs them most
The Pyramid of Pain: hashes and IPs are trivial to change; TTPs are tough.
Threat intelligence gives defenders clues to block. Not all clues are equal. At the bottom, the fingerprint of a malware file, its hash. Change one byte, and the hash is new. Trivial. An IP address: rent another server for a few dollars. A domain name: register a new one. Network and host artefacts, such as a particular file path, take a little effort. Tools: now the attacker must find or write new ones. At the top, tactics, techniques and procedures: how they work. Detect that, such as any program reading passwords from memory, and the attacker must learn new habits. That hurts.
Threat intelligence · Going deeper · 6 of 6
Where intelligence comes from, and how to use it
Threat intelligence turns raw indicators into decisions: what to block, hunt and patch first.
Threat intelligence turns raw indicators into decisions: what to block, hunt and patch first. Sources. Open source (OSINT) vendor and government advisories, security blogs, public reports, social media. CISA Known Exploited Vulnerabilities catalog flaws confirmed as exploited in the wild; patch these first. Sharing communities ISACs (information sharing and analysis centres) for each industry, such as FS-ISAC for finance. Commercial feeds curated indicators and reports on threat groups, delivered to firewalls and SIEMs. Dark web monitoring stolen credentials or data offered for sale that belong to you. Formats. Indicators are shared as STIX (a structured format for threat information) over TAXII (a protocol to exchange it), so tools can consume them automatically. Three levels of intelligence.
Sources
- Open source (OSINT) vendor and government advisories, security blogs, public reports, social media.
- CISA Known Exploited Vulnerabilities catalog flaws confirmed as exploited in the wild; patch these first.
- Sharing communities ISACs (information sharing and analysis centres) for each industry, such as FS-ISAC for finance.
- Commercial feeds curated indicators and reports on threat groups, delivered to firewalls and SIEMs.
- Dark web monitoring stolen credentials or data offered for sale that belong to you.
Formats
Indicators are shared as STIX (a structured format for threat information) over TAXII (a protocol to exchange it), so tools can consume them automatically.
Three levels of intelligence
Ransomware as a service
Ransomware is an industry, with suppliers, brokers and affiliates
In ransomware as a service, developers rent the kit, access brokers sell the way in, affiliates attack. Double extortion adds a leak threat.
The most damaging attackers today run like a company. Developers write the ransomware and rent it out for a share of each payment. Access brokers specialise in getting into companies, and sell those footholds. Affiliates buy both and run the attack. They keep most of the money. The victim’s data is copied first, then encrypted. Pay, or it is published. That second threat is why backups alone no longer end the crisis. In 2025, ransomware featured in forty-four percent of breaches. The median payment was a hundred and fifteen thousand dollars, and more victims than ever, sixty-four percent, refused to pay.
Supply chain attacks
One breached supplier opens the door to every one of its customers
A supply chain attack compromises a trusted supplier, update or service to reach its customers.
Why attack a thousand companies when you can attack the one they all trust? Third-party involvement in breaches doubled in a single year, from fifteen to thirty percent. One supplier is compromised, and its updates or its access carry the attacker into every customer at once. Target came in through a heating supplier. SolarWinds shipped a poisoned update that up to eighteen thousand customers installed. A flaw in the MOVEit file-transfer product exposed thousands of organisations. And in 2024 a backdoor slipped into XZ Utils, a compression library inside most Linux systems, was caught by one engineer before it spread.
AI-enabled threats
AI makes old attacks cheaper and opens new ones against AI itself
Attackers use AI for deepfakes and phishing at scale; AI systems face prompt injection, data poisoning and jailbreaks.
Artificial intelligence changes both sides. In early 2024, an employee in Hong Kong joined a video call with the company’s finance chief and several colleagues. Every one of them was a deepfake. The employee made fifteen transfers worth about twenty-five million dollars. Language models also write fluent, personal phishing in any language, at no cost. And AI systems are a target in their own right. Prompt injection hides instructions in content a model reads. Poisoning plants behaviour through training data. Jailbreaks talk a model past its rules. Invented package names become a trap when attackers publish them.
Lesson complete
You finished “Attackers and how attacks unfold”
Nicely done.