Frameworks Guidance from a public authority Bundesamt für Sicherheit in der Informationstechnik (BSI)
BSI guidance on generative AI
The German cybersecurity authority on generative AI models: risks from use, misuse and attacks, each with mitigations.
What this framework governs
The Federal Office for Information Security (BSI) is Germany’s federal cybersecurity authority and the point of reference for information security for companies and public administration in Germany.
Its paper on generative AI models, addressed to industry and public authorities, describes where such models carry risks in proper use, through misuse and through targeted attacks, and names mitigations for each risk.
It explicitly addresses companies and public bodies that intend to deploy generative AI, which makes it directly tailored to practical use.
Mappings in the threat catalogue
BSI guidance on generative AI. Publisher: Bundesamt für Sicherheit in der Informationstechnik (BSI). The Versatile AI Risk Assessment threat catalogue reaches 40 of its 52 threats through this framework, backed by one document at 28 locations. Mappings are documented for 40 of the 52 threats in the catalogue. The overview shows their distribution by topic group.
12 threats without a mapped reference
The catalogue contains no reference from this framework for these threats. This does not establish whether the original document addresses them. See the threat pages for recorded mappings to other frameworks.
- Cost Harvesting / Repurposing Attacks on the Running Model and Service
- Application Vulnerabilities Application and Integration Security
- Application Denial of Service Attacks on the Running Model and Service
- Controversial Topics Harmful Content
- Self-harm Harmful Content
- Unethical Actions Harmful Content
- Multimodal Attacks Prompt Attacks and Guardrail Evasion
- Middleware Exploits (AI Framework Attacks) Application and Integration Security
- Side-Channel Attacks (Timing Analysis) Attacks on the Running Model and Service
- Shadow AI (Unsanctioned AI Service Use) Reliability and Responsible Use
- Agent Memory Poisoning (Persistent Context) Agentic and Autonomous AI
- Insecure Inter-Agent Communication (A2A/MCP) Agentic and Autonomous AI
Analysis: combine several frameworks and see the gaps that remain
Threats referencing this framework (40)
Grouped by topic. Every entry leads to the full threat page.
Agentic and Autonomous AI
3 threatsApplication and Integration Security
5 threatsAttacks on the Running Model and Service
4 threatsHarmful Content
6 threatsMalicious Use for Attacks, Fraud and Disinformation
4 threatsModel and Training Data Manipulation
4 threatsPrivacy and Data Leakage
4 threatsPrompt Attacks and Guardrail Evasion
4 threatsReliability and Responsible Use
3 threatsSupply Chain and Provenance
3 threatsNo threat matches this input. Reset the filters to see all of them again.
Sources used (1)
These exact versions are pinned in the catalogue. Every page states the publisher, the version used, the locations and the referencing threats.
This page does not reproduce the text of the standards. It states the identifier, title and location; the wording itself is in the original document. The mappings are taxonomic and not evidence of compliance.
The guidance provides technical recommendations and is not itself a legal norm. Which measures are binding for an organisation must be assessed against the applicable requirements.
Further frameworks
The frameworks address different aspects: legal requirements, organisational risk management and technical security. Explore the further perspectives included in the catalogue.
From the framework to the assessment
The full catalogue states the mitigations, the possible impact and every verified location for each threat. The live demo runs locally in your browser, with no sign-up.
Cite this page
For reports, policies or internal documents; the link leads directly to this framework page.
“BSI guidance on generative AI” (Bundesamt für Sicherheit in der Informationstechnik (BSI)). Mappings in the AI threat catalogue, Versatile AI Risk Assessment, as of July 2026. https://www.versatile-ai-risk-assessment.com/en/wissensbasis/frameworks/bsi-generative-ai/