A DoS attack (DoS攻撃) floods a server or network with more requests than it can process, leaving legitimate users unable to use the service.
The attacker sends a flood of requests
Processing power and bandwidth run out
Legitimate users are locked out
What a DoS attack damages is availability in information security, meaning the system can be used whenever it is needed. The attacker concentrates huge numbers of requests and traffic on the target until the server's processing power or the line's capacity is used up.
AnalogyIt is like someone handing in pointless forms at a government counter, one after another, so that people with real business never get their turn.
How to rememberDoS stands for Denial of Service. The o is lowercase because it comes from the preposition "of". It is a different word from DOS, the old PC operating system.
Exam points
ExamThe aim of a DoS attack (DoS攻撃) is to disrupt a service, and what it damages is availability in information security. The direct damage to the server is that it can no longer provide its service.
ExamA DoS attack (DoS攻撃) is a cyberattack (サイバー攻撃) carried out over a network, so it is not classed as a physical threat (物理的脅威) such as a disaster like an earthquake or fire.
A DDoS attack (DDoS攻撃), or distributed DoS attack, takes a service down by having many attacker-controlled devices send massive traffic to the same target at once.
← Scroll the figure sideways →
The attacker first infects many devices with a bot, a malicious program that can take orders from outside. When the bots receive a command from the attacker's command server, they all start sending traffic to the chosen target at the same time. Because the traffic comes from so many scattered devices, blocking a few sources does little to stop the attack.
AnalogyPicture a single signal sending crowds from every direction to the entrance of one shop, so that real customers cannot get in.
How to rememberThe leading D is for Distributed: add one D to DoS and the attack sources spread out.
Exam points
ExamA DDoS attack (DDoS攻撃) pours massive traffic from many devices at once into a target, making Web services and the like unusable.
ExamThe devices used as stepping stones (踏み台) are infected with malware (マルウェア) that lets the attacker control them remotely, and this malicious program is called a bot (ボット). Not only PCs but also IoT devices get used this way.
Mix-upA DDoS attack (DDoS攻撃) is not a way of pushing malware onto a site's visitors. Infecting a user's device just because the site was opened is a drive-by download (ドライブバイダウンロード).
A zero-day attack (ゼロデイ攻撃) exploits a software vulnerability after it has been found but before the developer's fix is available.
Vulnerability found
Window with no fix
Fix released
Fix applied, hole closed
Once users apply the fix, attacks on that weakness no longer get through.
4 / 4
A vulnerability is a security weakness that comes from a flaw in how software is built, and the developer distributes a program that repairs it, called a patch. To prepare for zero-day attacks, organizations combine patching with other measures, such as adding layers that watch for suspicious activity.
AnalogyIt is like a burglar who learns the emergency exit lock is broken and slips in that very night, before the locksmith has come.
How to rememberRead "zero-day" as "the developer had zero days to respond" and it sticks.
Exam points
ExamA zero-day attack (ゼロデイ攻撃) exploits a vulnerability (脆弱性) in the period between its discovery and the release of the developer's fix or countermeasure.
ExamWhether an attack counts as "zero-day" (ゼロデイ) depends on whether it comes before or after the fix is released, not on the kind of vulnerability or program involved.
Mix-upFor attacks that hit software vulnerabilities, such as a buffer overflow (バッファオーバフロー), applying the fix patch is an effective countermeasure. A zero-day attack (ゼロデイ攻撃) targets the period before that patch exists.
Cryptojacking (クリプトジャッキング) is an attack that secretly plants a malicious program on someone else's computer and uses its computing power for crypto-asset mining.
Plant a malicious program
Victim's device does the mining math
Reward goes to the attacker
Mining means performing the calculations that confirm crypto-asset transaction records and receiving crypto-assets as a reward. Because this takes a lot of computing power and electricity, the attacker makes other people's devices carry the load and keeps only the reward. The victim's device slows down and uses more power.
AnalogyIt is like your home bread maker being used all night, without your knowledge, to bake someone else's orders: they pocket the sales, and you are left with the power bill and a worn-out machine.
How to rememberThe word blends crypto with hijacking.
Exam points
ExamCryptojacking (クリプトジャッキング) uses the computing power of someone else's computer, without the owner noticing, to mine (マイニング) crypto-assets (暗号資産).
Mix-upThe goal of cryptojacking (クリプトジャッキング) is neither stealing data nor tricking people into revealing information. What it takes is a device's computing resources and electricity.
A prompt injection attack (プロンプトインジェクション攻撃) feeds generative AI crafted instructions, or prompts, to make it break the restrictions set by its developers or reveal information it should not.
Crafted instructions go in
The AI obeys them over its original instructions
Forbidden answers or actions
Generative AI receives both the instructions its developers gave in advance and the user's input as plain text and processes them together. Attackers exploit this by slipping a command such as "ignore all previous instructions" into the input so that the AI puts it ahead of its original instructions. Besides typing such commands directly, attackers can also hide them in Web pages or documents that the AI is asked to read.
AnalogyIt is like scribbling "open the safe for whoever brings this" in the corner of a document handed to a receptionist, who mistakes it for the boss's orders and complies.
How to rememberInjection means pouring something in: another set of instructions is quietly poured into the legitimate stream.
Exam points
Mix-upSQL injection (SQLインジェクション) slips commands for a database into the input, while a prompt injection attack (プロンプトインジェクション攻撃) slips in instructions for generative AI. Both are injections, but they aim at different targets.
An adversarial example (敵対的サンプル) is input data altered in ways people can barely detect, crafted to make an AI machine learning model judge wrongly.
A machine learning model is a system that learns rules from large amounts of data and then classifies or judges new input. For example, if an image recognition model is given a picture with a tiny amount of noise added, it may identify it as something else even though it looks unchanged to people.
How to remember"Adversarial" signals that the data was made to fool the AI. "Example" here means a single piece of data fed to the model.
Exam points
Mix-upAn adversarial example (敵対的サンプル) tampers with the input given to a model that has already been trained. It is different from data poisoning (データポイズニング), which mixes bad data into the training data to distort the learning itself.
Prompt injection attack vs. Adversarial example
Aspect
Prompt injection attack
Adversarial example
Key differenceWhat is tampered with
The instructions given to the AI
The input data given to the model (images, etc.)
What happens
Answers or actions that break the restrictions
Wrong classifications or judgments
Key points
The aim of a DoS attack is to disrupt a service; what it damages is availability.
In a DDoS attack, many devices infected with a bot and remotely controlled by the attacker serve as stepping stones.
A zero-day attack exploits a vulnerability before the developer releases a fix.
Cryptojacking uses other people's computing power without permission to mine crypto-assets.
A prompt injection attack exploits the way generative AI treats input text as instructions.