Skip to content

GCRI · 人工知能と自律システム

AI のサイバーセキュリティ

AI and cybersecurity meet in three ways: AI システム can be attacked, AI can help attackers, and AI can help defenders. Models can be poisoned, stolen, manipulated or made to leak data; their supply chains of data, weights and libraries carry new risks. This page sets out how these threats work, how AI systems are secured, and what evidence serves the security teams and regulators responsible.

01登録

AI cyber security on the record.

02質問

AI cyber security: how the risk develops.

Adversarial methods and model exposure. The dossier below sets out how the risk develops, what is measured, where the data come from, how it is modelled and who decides.

メカニズム

Models are attack surfaces

Adversarial inputs, prompt injection, data extraction and model theft target the model itself. Classic application vulnerabilities still apply around it.

メカニズム

The AI supply chain is long

Training data, pretrained weights, libraries and plugins come from many sources. Poisoned data or tampered models can introduce hidden behaviour.

メカニズム

AI lowers the cost of some attacks

Generative AI can draft phishing, translate lures and help write code, lowering the skill barrier for some attacks. The size of this effect is studied and debated.

メカニズム

AI assists defence

Machine learning detects anomalies, triages alerts and finds vulnerabilities. Defenders must also secure the AI tools they adopt.

03シミュレーター

相互依存するネットワークを通る連鎖。

A ハザードの影響範囲 は各層で同じ場所にある資産を奪います。その後、断片化、層間の依存、生き残った資産の過負荷が障害を広げます。3D ビューは層を奥行きで分けて示します。例示的な値を明記した教材用モデルであり、予測ではありません。

管理策

レイヤーごとの資産
18
ハザードの影響範囲
面積の 8%
別のレイヤーに依存する資産
60%
予備容量(許容度)
30%

読み取り値

ハザードを受けた資産
7%
稼働を続ける資産
50%
初期被害の増幅
6.8 times
決着までのステップ数
3
数値、手法、出典
ステップ稼働中のデータセンター(%)稼働中のネットワーク(%)稼働中の電力(%)
08394100
1727894
2616750
3616722

手法

  1. 相互依存ネットワーク Buldyrev ほか(2010)、Catastrophic cascade of failures in interdependent networks、Nature 464
  2. 過負荷カスケード Motter and Lai (2002), Physical Review E 66
  3. 媒介負荷 Brandes (2001), Journal of Mathematical Sociology 25
  4. 多層ネットワーク Kivela and others (2014), Journal of Complex Networks 2
  5. インフラの相互依存 Rinaldi, Peerenboom and Kelly (2001), IEEE Control Systems 21(6)
  6. 不確実性ファン(パラメータ・モンテカルロ) Spiegelhalter, Pearson and Short (2011)、Visualising uncertainty about the future、Science 333
  7. 一因子ずつの感度分析(トルネード図) Saltelli and others (2008), Global Sensitivity Analysis: The Primer, Wiley

データ

The model runs on the teaching values set by the controls. To calibrate it for a place, read it against these admitted layers and series, which this page’s map can draw:

  • 主要道路 Natural Earth
  • 鉄道 Natural Earth
  • 港湾 Natural Earth
  • 空港 Natural Earth
  • 都市域 Natural Earth
  • 今夜の夜間光と停電(VIIRS 昼夜バンド) NASA/NOAA VIIRS(Suomi NPP)· 登録 nasa.gibs.context-live-layers
  • 自然現象(NASA EONET) NASA EONET · 登録 nasa.eonet.events
相互依存するネットワークを通じた連鎖。説明用の教材モデルで、実測データではなくコントロールで設定した値に基づいて動きます。手法:相互依存ネットワーク(Buldyrev and others (2010)、Catastrophic cascade of failures in interdependent networks、Nature 464)、過負荷の連鎖(Motter and Lai (2002)、Physical Review E 66)、媒介負荷(Brandes (2001)、Journal of Mathematical Sociology 25)。
04分野を横断して

AI changes cyber risk in every domain’s control systems.

05手法とエビデンス

How AI cyber security is modelled, measured and tested.

モデル、その仮定、データ、不確実性を、科学者やエンジニアが確認する形で示します。構造は公表された手法に従っています。描かれた値は例示であり、知見ではありません。

システム・モデル

The couplings between the systems in AI cyber security: signs of influence, a reinforcing and a balancing loop, and cross-links where stress jumps.

++−+人工知能データコミュニティ経済R 自己強化B バランス
Causal loop diagram from the page’s systems; signs show the direction of influence. Illustrative structure for a system dynamics or network model, not estimated couplings.
FAIR 原則・GO FAIR と FORCE11W3C PROV・World Wide Web Consortium不確実性の一貫した扱いに関するガイダンス・IPCCCEOS 解析即応データ・地球観測衛星委員会
06指標

何を、どの程度うまく測定しているか。

各指標には、その単位、更新頻度、背後にある出典の種類、値が持つことになるエビデンスのラベルが付いています。値は表示しません:プログラムが自らの権限のもとで値を提供します。

指標単位周期出典の種類ラベル
Vulnerabilities reported in AI frameworks and librariescount of CVE entries月次Vulnerability databases報告値
Adversarial robustnessaccuracy under a stated attackper evaluationRobustness testing観測
Models with signed provenancepercent of deployed modelsper inventoryModel registries申告済み
AI-related security incidentscount by technique年次Incident reports報告値
Time to detect AI system compromisehoursper incidentIncident response records観測
Organisations with AI security policiespercent of surveyed organisationsper surveyIndustry surveys報告値
07出典

エビデンスの出所。

名前の明示された公開データセットと機関の記録。それぞれにライセンス区分があります。合意に基づく出典は、その条件のもとでのみ使用し、再公開することはありません。

出典提供するものライセンス区分
MITRE ATLASAdversarial tactics and techniques against AI システム, with case studiesPublic, licence to check
NIST adversarial machine learning taxonomyTerminology and taxonomy of attacks and mitigationsPublic domain, US government
National Vulnerability Database and CVE recordsVulnerabilities in AI softwarePublic domain, US government
ENISA reports on AI cybersecurityThreat landscape and good practice for securing AIPublic, ENISA terms
Joint guidance from national cyber agencies on secure AI system developmentSecure design, development, deployment and operationPublic, terms per agency
08決定

誰が決め、エビデンスが何に役立つか。

人工知能と自律 システム prepares evidence. The decisions below are taken by the authorities named, under their own rules, and never by the platform.

誰が決定するか質問このページが提供するエビデンス
Chief information security officersHow to secure AI systems the organisation builds or buysThreat models; supply chain checklists
National cybersecurity agenciesWhat guidance to issue on AI securityIncident and vulnerability evidence
開発者Which defences to build into models and applicationsRobustness testing; attack taxonomies

このトピックが生み出す成果物

  • Threat models for AI deployments
  • AI supply chain security checklists
  • Robustness testing protocols
  • Evidence notes on AI-enabled attack trends
09中核となる問い

What security researchers ask.

  • How can inputs be crafted to make a model misbehave, and how robust are defences?
  • Can an attacker poison training data or tamper with model weights in the supply chain?
  • Can private training data or a model itself be extracted through its outputs?
  • How does AI change attackers’ and defenders’ capabilities?
10メカニズム

The AI attack surface.

Adversarial examples are inputs altered in ways imperceptible to people that cause models to err. Data poisoning inserts malicious examples into training data to plant backdoors. Membership inference and extraction attacks recover information about training data or replicate a model through repeated queries. For language models, jailbreaks and prompt injection bypass intended restrictions.

On the other side, AI can help attackers find vulnerabilities, write malware variants and craft convincing phishing at scale, and help defenders triage alerts, analyse code and detect anomalies. The balance between the two is an open research question.

11手法とデータ

Security engineering for AI.

  1. 01

    Red-teaming

    Structured attempts to break a system, combining サイバーセキュリティと暗号 and artificial intelligence and machine learning expertise. Red-teaming finds problems but cannot prove their absence.

  2. 02

    脅威モデリング

    Mapping assets, adversaries and attack paths across data, training, deployment and supply chain.

  3. 03

    Robustness testing

    Measuring performance under adversarial perturbation. Robustness to known attacks does not imply robustness to new ones.

  4. 04

    Supply-chain integrity

    Provenance of datasets, models and dependencies, with signing and verification. Provenance tooling is still maturing.

13Where this page’s references

Where this page’s references lead.

このページの読解の基準となる機関と文書。状態と版は変わります。各リンクは発行機関へ移動します。

14質問

質問と回答。

What is data poisoning?
Manipulating training data so that a model learns hidden behaviours or performs worse, for example responding to a trigger in a way the attacker chooses.
What is MITRE ATLAS?
A knowledge base of adversary tactics and techniques against AI systems, modelled on MITRE ATT&CK, with real-world case studies.
Does AI make cyberattacks easier?
It can lower the effort for some tasks, such as writing phishing messages, according to agency assessments. How much it changes overall attack capability is still being studied.
Does the platform publish attack techniques?
Only at a level that supports defence, following responsible disclosure. Details that would mainly enable harm are not published.
How do I report a vulnerability?
宛先 セキュリティに関する報告フォームに送ることができる。受領通知には参照番号と次のステップが示される。
Is open release of model weights a security risk?
It is a trade-off. Open weights enable scrutiny, research and local control, and also make misuse and the removal of safeguards easier. The balance depends on the model’s capabilities.
Can AI defend against AI-enabled attacks?
It can help, but it is not a complete answer. Basic security practices, monitoring and human expertise remain essential.
報告書貢献する

このページからエビデンスを提供する。

報告書 is the Consortium’s quarterly publication. What you know about this subject can become evidence, a case or a correction in the next edition, credited to you and published under CC BY 4.0 where it is your own work.

GCRI人工知能と自律システム指数関数的テクノロジーLabs のエビデンス
人工知能と自律システム

手法、データセット、または問いを持ち込んでください。

情報提供と問い合わせは本日から受け付けています。問い合わせには参照番号が付き、次の手順が示されます。申請、支払い、計算資源の割り当て、登録は、担当チーム、規約、審査体制が整ってから開始します。