UMUYOBOZI W'umuryango

Umutekano wa AI

AI security protects models, data, tools, and surrounding services from unauthorized access or manipulation.

2 min somaIbiherutse kuvugururwa Part of the AI Policy & Society learning path

Incamake

It includes ordinary software security and threats that target learning or model behavior. A secure design begins with the assets, adversaries, and trust boundaries of the actual application.

Ibyingenzi byingenzi

  • Threat-model the full application.
  • Enforce permissions outside the model.
  • Retest controls across system changes.

Kwibira cyane

Identify what needs protection: private inputs, training data, model artifacts, credentials, connected accounts, and external actions. Record who can influence each input and what an attacker could gain from a failure. A public chatbot and an internal agent with write access have different threat models. Threats can affect different stages. Poisoned training material can alter learned behavior; adversarial inputs can manipulate predictions; untrusted retrieved content can redirect a tool-using application. Model output can also become dangerous when inserted into a database query, webpage, or command without appropriate handling. Apply controls at the software boundary. Enforce authorization in code, keep secrets out of model-visible context where possible, restrict tool scope, and validate outputs before use. A prompt asking a model to behave safely cannot replace account isolation or permission checks. Test representative failure paths in an authorized environment and maintain an incident process. Log enough information to investigate without collecting unnecessary sensitive content. Evaluate controls after changes to the model, retrieval sources, tools, and dependencies. Describe residual risk honestly; no single filter establishes complete protection.

Ubushishozi

A model refusing one malicious prompt does not prove that a system is secure. Different inputs, tools, modalities, and component boundaries can create distinct failure paths.

Locate the security boundary

  1. Imagine an assistant searching a private document store for a signed-in user.
  2. Apply the user’s access filter in the retrieval service before documents enter the model context.
  3. Test with a document belonging to a different account and verify that neither its contents nor identifying metadata appear in the result.

This defensive, hypothetical test checks authorization independently of the model’s willingness to follow instructions.

Ingaruka z'Ingamba

Risk and safety

Catastrophique na burimunsi AI yangiza byombi biterwa nuwumva ingaruka ninde ushobora gukora.

Ibyemezo bisobanutse

Kumenya gusoma no kwandika rusange kandi byumwuga byerekana niba politiki yumutekano ikomeye ishoboka muri politiki.

Cutting through hype

Ibisobanuro bisobanutse bigabanya gufatwa ukoresheje impuha, laboratoire PR, hamwe namakinamico adasobanutse.

Gushyira mu bikorwa Isi

Check that one account cannot retrieve another account’s documents.

Validate generated fields before using them in a database operation.

Ingaruka & Kurinda

Gufata ibyago bibaho nka sci-fi mugihe ubushobozi bwimbaraga.

Kwitiranya umutekano wibicuruzwa byo hejuru hamwe no guhuza munsi y'ubwigenge buhanitse.

Kureka abatari Icyongereza nabatari abahanga bafite isoko yo hasi gusa.

Igishushanyo mbonera

1

Gutandukanya ibicuruzwa byangiza, gukoresha nabi, no gutakaza-kugenzura / ingaruka mbi.

2

Baza ibimenyetso byahindura uko ubona ku gihe n'uburemere.

3

Hitamo inkomoko yibanze nibisobanuro bifatika kubisabwa byo kwamamaza.

4

Menya inzira imwe y'ibikorwa: umwuga, politiki, inkunga, cyangwa ubuhanga - ntabwo ari ukumenya gusa.

Inkomoko no gusoma

Komeza Ubushakashatsi

Free newsletter

Get the daily AI briefing

Three verified AI stories every weekday morning, written in plain English. Free forever, no ads.

One email each weekday. Unsubscribe in one click. We never sell or share your address.

Test yourself

Take the AI Security quiz

Instant feedback on every answer, and a shareable certificate with a verifiable ID once you pass a course.

Tangira ikibazo

Support free AI education. AI Understanding is a 501(c)(3) nonprofit — no ads, no paywall, ever. Make a donation

Next in AI Policy & Society

Umutekano wa AI

Ibibazo bikunze kubazwa

Is a strong system prompt enough to secure an assistant?

No. Authentication, authorization, input and output handling, tool limits, and incident response remain necessary parts of the application.