Secure, compliant and trustworthy AI
Security testing for LLM applications and AI agents, AI governance and practical support for the EU AI Act, so you can use AI with confidence.
Overview
Use AI without opening new doors to attackers.
- LLM and AI agent security testing
- EU AI Act readiness
- OWASP Top 10 for LLM, NIST AI RMF, ISO/IEC 42001
AI brings new risks: prompt injection, leaks of confidential data, agents taking actions nobody intended and tools adopted without control. At the same time the EU AI Act sets new rules: transparency obligations apply since 2 August 2026, and requirements for high-risk systems follow by December 2027.
We combine AI engineering with offensive security. We find out which AI systems your company uses, test them like an attacker would, add the safeguards they need and set up the governance that keeps them compliant over time.
What we do
AI security and governance services
01
AI inventory & classification
A complete list of the AI systems you build or use, including AI inside purchased software, classified by risk under the AI Act.
02
LLM penetration testing
Red teaming of chatbots, assistants and agents: prompt injection, jailbreaks, data leakage and abuse of connected tools.
03
Agent guardrails
Least-privilege tool access, approval steps, sandboxing and limits that keep agents within their intended scope.
04
Data protection for AI
Controls that keep personal and confidential data out of prompts, logs and models, aligned with GDPR.
05
AI Act transparency
AI disclosure for chatbots and voice assistants, and machine-readable marking of AI-generated content.
06
Governance framework
Policies, roles, approval processes and documentation for responsible AI, aligned with ISO/IEC 42001.
07
Monitoring & audit trails
Logging of prompts, answers and agent actions, with alerts for anomalies and evidence for audits.
08
AI literacy training
Training for management and staff, with the records required by Article 4 of the AI Act.
Why SetExe
What secure, compliant AI brings you
Fewer surprises
Vulnerabilities in AI applications found and fixed before they reach customers or attackers.
Audit-ready
Inventory, policies, logs and training records that show how you manage AI risk.
Faster approvals
New AI projects pass security, privacy and legal reviews more quickly with a clear framework.
One team for AI and security
The engineers who test your AI can also build and fix it.
How we work
How we secure your AI
Phase 01
Inventory
Discovery of the AI systems in use and of the data and tools they can access.
Phase 02
Assess
Risk classification and security testing of the most exposed systems.
Phase 03
Protect
Guardrails, access control, data protection and AI disclosure implemented.
Phase 04
Govern
Policies, roles, training and documentation for ongoing compliance.
Phase 05
Monitor
Continuous logging, periodic re-testing and updates as rules evolve.
Technologies
AI security tools and standards
The platforms, frameworks and standards we use most often for this service.
- OWASP Top 10 for LLM
- OWASP GenAI Security Project
- MITRE ATLAS
- NIST AI RMF
- ISO/IEC 42001
- EU AI Act
- GDPR
- Promptfoo
- PyRIT
- garak
- Llama Guard
- Azure AI Content Safety
Ways to work together
Fixed-scope assessment
Defined scope, timeline and price, with a clear report and priorities.
Ongoing programme
A team that implements and improves the measures month by month.
Experts on demand
Specialists who join your team when you need them.
FAQ
Questions about AI Security & Governance
Does the EU AI Act apply to our company?
It applies to companies that develop or use AI systems in the EU, with obligations that depend on your role and on the risk level. Transparency rules apply since 2 August 2026; most high-risk requirements apply from 2 December 2027.
What is prompt injection?
An attack that hides instructions in text, emails or web pages so that an AI system ignores its rules, reveals data or performs unwanted actions. It is the top risk in the OWASP Top 10 for LLM applications.
Can you test AI tools we bought rather than built?
Yes. We review the configuration, permissions and data flows of purchased AI tools and assistants, as well as the systems you develop.
Do you replace our lawyers?
No. We cover the technical and organisational side and work together with your legal advisers or our partner law firms on the legal interpretation.
What is AI literacy training?
The AI Act asks companies to support the AI skills of the people who use AI systems. We provide practical training and the records that document it.
SetExe provides technical and organisational services. Legal advice on the EU AI Act is provided by your legal advisers or by our partner law firms.
Related services
Services that complement AI security
Cybersecurity
Security assessments, penetration testing, secure development and cloud protection that reduce your risk and help you meet NIS2, GDPR and ISO/IEC 27001 requirements.
AI Agents & Automation
Assistants and agents that read your documents, answer your customers and automate repetitive processes, connected to your systems and governed with clear controls.
NIS2 Compliance
Gap analysis, security measures, incident reporting and documentation to meet NIS2 and Italy’s Legislative Decree 138/2024, delivered by engineers who can also implement the fixes.
Contact
Talk to an AI security expert
Share a few details and we will get back to you with next steps.
What happens next
- 1 · Tell us about your needs with the form
- 2 · Free consultation with one of our experts
- 3 · Proposal with scope, timeline and estimate