--- title: AI at avantguard · avantguard cyber security description: "How avantguard uses AI in penetration tests and red teaming: for specific subtasks, with local models for sensitive data and under clear rules." --- [Skip to content](https://avantguard.io/en/ai-at-avantguard#main) [![avantguard cyber security](https://avantguard.io/hs-fs/hubfs/LogoInverted.png?width=128&height=61&name=LogoInverted.png)](https://avantguard.io/en/?hsLang=en) - [Services](https://avantguard.io/en/services?hsLang=en) - [About Us](https://avantguard.io/en/about-us?hsLang=en) - [Experience](https://avantguard.io/en/experience?hsLang=en) - [Career](https://avantguard.io/en/careers?hsLang=en) - [Research](https://avantguard.io/en/blog?hsLang=en) - [Contact](https://avantguard.io/en/contact?hsLang=en) [DE](https://avantguard.io/de-ch/ai-einsatz) EN [Request an offer](https://avantguard.io/en/request-an-offer?hsLang=en) [DE](https://avantguard.io/de-ch/ai-einsatz) EN [Request an offer](https://avantguard.io/en/request-an-offer?hsLang=en) AI at avantguard # Attacking realistically today means: with AI. Attackers use AI and gain enormous efficiency from it. To keep your attack simulation true to life, it uses the same means: following recognised best practices and with local AI models for particularly sensitive information. Why AI in your test ## The threat is changing. Your test keeps pace. Attackers use AI along the entire attack chain: for reconnaissance, for phishing and for developing their tools. A realistic security test has to reflect this level. That is why AI is part of your test too: for specific subtasks and under the direction of experienced testers. ### More speed The time between the disclosure of a vulnerability and the first attack keeps getting shorter. Attack tools are now built in hours instead of days. ### More reach AI analyses public information about your company at scale and turns it into personalised phishing emails in any language. ### More adaptability Attack tools are modified quickly until security controls no longer detect them. Use by sensitivity ## Frontier model or local AI: what matters is what is at stake. Which model is used is decided per task, together with you. The deciding factor is how sensitive the information is. Typical examples from practice: ### Frontier models The most capable cloud models from leading providers. For tasks without particularly sensitive client information. - OSINT - Non-sensitive code reviews - Phishing - Security research ### Local AI models Sensitive data Run on in-house infrastructure in Rapperswil-Jona, Switzerland. Your data never leaves this environment. - Sensitive code reviews - Reverse engineering - Analysis of client documentation - Exploit development Safe use ## Clear limits for every use of AI. The basis is recognised best practice for using AI in offensive security, for example [Anthropic's recommendations for the safe use of AI](https://support.claude.com/en/articles/17317514-agent-containment-best-practices-for-cvp-participants). This applies to every engagement: ### Sensitive information stays in-house As a rule, AI processes sensitive data on company-owned hardware. Frontier models are only used for sensitive data at your explicit request. ### Humans in control Experienced penetration testers carry out the engagement. AI only takes on individual subtasks, clearly defined and strictly controlled. Every finding is verified personally, so you can rely on every result. ### Isolated environment The AI tools run in isolated environments. Outbound connections are restricted to what is necessary. ### No autonomy on critical systems AI never acts on critical production systems. Tests there run on replicas or read-only. ### Clear scope per engagement Permitted targets, actions and network boundaries are defined before every use. ### Traceable logs Every use of AI is logged and remains traceable for you. AI & LLM Security Testing ## Do you use AI yourself? Chatbots, AI agents and AI integrations create new attack surfaces. Have them tested before someone else does. [Get in touch](https://avantguard.io/en/contact?hsLang=en) ## Questions? Find out how AI is used in your project. [+41 55 253 30 30](tel:+41552533030) [info@avantguard.io](mailto:info@avantguard.io) [![avantguard cyber security](https://avantguard.io/hs-fs/hubfs/LogoInverted.png?width=140&height=66&name=LogoInverted.png)](https://avantguard.io/en/?hsLang=en) avantguard cyber security AG Neue Jonastrasse 52 8640 Rapperswil-Jona, Switzerland [+41 55 253 30 30](tel:+41552533030) [info@avantguard.io](mailto:info@avantguard.io) ## Company - [Services](https://avantguard.io/en/services?hsLang=en) - [About us](https://avantguard.io/en/about-us?hsLang=en) - [Experience](https://avantguard.io/en/experience?hsLang=en) - [AI at avantguard](https://avantguard.io/en/ai-at-avantguard) - [Careers](https://avantguard.io/en/careers?hsLang=en) - [Research](https://avantguard.io/en/blog?hsLang=en) - [Contact](https://avantguard.io/en/contact?hsLang=en) ## Services - [Penetration Testing](https://avantguard.io/en/penetration-test?hsLang=en) - [Red Teaming](https://avantguard.io/en/red-teaming?hsLang=en) - [Attack Simulation](https://avantguard.io/en/attack-simulation?hsLang=en) - [FINMA-regulated pentesting](https://avantguard.io/en/finma-regulated-pentesting?hsLang=en) - [AD Security Improvement](https://avantguard.io/en/ad_improvement?hsLang=en) - [mesher](https://avantguard.io/en/mesher?hsLang=en) - [All services](https://avantguard.io/en/services?hsLang=en) ## Research - [The Blind Spot Of 802.1X Network Access Control (NAC)](https://avantguard.io/en/blog/die-blinde-stelle-von-802.1x-network-access-control-nac?hsLang=en) - [BlueHammer and RedSun analysed](https://avantguard.io/en/blog/bluehammer-und-redsun-analysiert?hsLang=en) - [Plaintext in Memory: Revisiting Browser Password Security](https://avantguard.io/en/blog/klartext-im-speicher-neubewertung-der-passwortsicherheit-im-browser?hsLang=en) - [Experience Report: AI-Assisted BOF Development in Red ...](https://avantguard.io/en/blog/erfahrungsbericht-ki-gest%C3%BCtzte-bof-entwicklung-im-red-team?hsLang=en) - [All posts](https://avantguard.io/en/blog?hsLang=en) © 2026 avantguard cyber security AG - [Privacy policy](https://avantguard.io/en/privacy-policy) - [Deutsch](https://avantguard.io/)