Clear rules for the AI tools your staff already use, assessed risks for the higher-impact uses and hands-on testing of the AI systems you run, so your team can use AI with less risk of leaking data.
It suits organisations whose staff already use AI, or that run their own AI systems, and need clear rules, assessed risks and evidence they can show a board, customer or auditor.
Built for
Organisations with 5 or more staff, including not-for-profits
Boards, executives and the people who own risk
Government, health and education teams
Teams that build or run systems on large language models (LLMs)
You probably need it if…
Staff use ChatGPT, Copilot or other AI tools without a written policy
Your board needs to show accountability for AI, including under ASD Essentials
You must assess AI risk before procurement or deployment
You run a chatbot, AI agent or document search tool built on a large language model
You hold ISO/IEC 27001 and want to add ISO/IEC 42001
02What's included
What you get
Govern
AI acceptable use policy
A plain-English AI policy for staff covering which tools are approved, what data must never be entered and who signs off on new AI use. It sits alongside your existing information security policies rather than in a separate binder.
Shadow AI discovery and inventory
We check your Microsoft 365 and network logs and run a short staff survey to see which AI tools are in use. Each one goes into an AI inventory with an owner, the data it touches and a risk rating.
AI risk register and impact assessments
Each AI use is assessed for privacy, security, accuracy and fairness risks, with treatments agreed and recorded. Higher-impact uses get a documented impact assessment you can show an auditor, customer or board.
Review & test
Third-party AI vendor review
We review the AI features inside software you already pay for, and new AI vendors before you sign. You learn where your data goes, whether it is used to train models, where it is stored and what the contract allows.
LLM and AI agent security testing
Hands-on testing of your chatbot, AI agent or document search tool against the OWASP Top 10 for LLM Applications. Findings come with evidence and practical fixes.
Standards readiness
ISO/IEC 42001 readiness
A gap assessment against ISO/IEC 42001:2023 and a roadmap that reuses your ISO/IEC 27001:2022 management system where the two standards overlap, so you don't end up running two separate systems.
Secure AI
Getting agreement on which AI tools staff may use
An AI policy holds when the people using the tools helped shape it. We run working sessions with leadership and the teams already experimenting, so the approved list reflects real tasks and the rules are clear.
03In detail
AI security & governance, in depth
What does AI governance look like for a 30-person business?
It should fit on a few pages and be light enough to keep current. For around 30 staff, we recommend:
A one-page AI policy covering approved uses, banned data and who approves new tools.
An approved-tools list with the business account each tool must be used through.
Discovery of the AI tools already in use, repeated regularly.
A regular review of the tools list, the risk register and any incidents.
One accountable owner, usually an executive, who approves new AI uses.
Each AI use then gets a line in a simple risk register:
Use
Main risk
Treatment
Owner
Drafting client emails with a business AI account
Client personal information entered into the tool
Business account with training switched off, staff briefing, no tax file numbers or health details
Operations manager
Which AI frameworks should an Australian business follow?
You don’t need every framework. We use each for what it does well and map your controls across them, so one piece of work supports several obligations.
Six essential practices: decide who is accountable, understand impacts and plan accordingly, measure and manage risks, share essential information, test and monitor, and maintain human control
Sets the security risks we test for in chatbots, agents and document search tools
The Guidance for AI Adoption is the current Australian guidance. Published in October 2025, it supersedes and builds on its predecessor, Australia’s Voluntary AI Safety Standard, and its ten guardrails. If you have no management system yet, our ISO 27001 consulting service can set up the foundation that ISO/IEC 42001 builds on.
The Australian Signals Directorate (ASD) has also released the ASD Essentials series, which includes an “Executive artificial intelligence accountability” principle. It sets out what executives are accountable for when their organisation uses AI, and ASD treats AI as an asset to include when scoping the Essentials domains. Our ASD Essentials page explains the series.
The Privacy and Other Legislation Amendment Act 2024 also added automated decision-making transparency obligations. Once they commence, privacy policies will need to explain when a computer program uses personal information to make a decision, or to do something substantially and directly related to making it, where the decision could significantly affect someone’s rights or interests. The OAIC has published guidance on these transparency obligations, and we help you identify which of your processes they cover.
How do we test AI systems for security weaknesses?
Any AI system that reads untrusted text can be attacked through that text. Our penetration testing team tests these systems hands-on, guided by the OWASP Top 10 for LLM Applications and the MITRE ATLAS knowledge base of attacks on AI systems. For agents, we also use the OWASP Top 10 for Agentic Applications 2026. The main risks we look for are:
Prompt injection (LLM01): we try direct attacks and instructions hidden in emails, web pages and uploaded files.
Sensitive information disclosure (LLM02): the system revealing personal data or documents a user should not see.
Excessive agency (LLM06): an agent holding more permissions or tools than its job needs.
System prompt leakage (LLM07): the hidden instructions, and any credentials or business rules in them, being exposed.
Vector and embedding weaknesses (LLM08): document search that skips access checks or can be poisoned.
Each finding is reported in a consistent format. For example: “Indirect prompt injection through uploaded PDF. Risk: high. Evidence: a hidden instruction in a test document made the assistant email its summary to an external address. Fix: require approval for outbound email and strip hidden text before processing.”
04The process
How it works
1Step 01
Discover
We interview key people, review your current policies and run shadow AI discovery to see how AI is really used today.
2Step 02
Assess
Each AI use is mapped to the NIST AI RMF, ISO/IEC 42001, the Guidance for AI Adoption and your legal obligations, and each risk is rated.
3Step 03
Govern
We write or update your AI policy, risk register and approval process, and brief executives on their accountability for AI.
4Step 04
Test
Where you run your own AI systems, we test them hands-on and report each finding with a fix.
5Step 05
Review
We agree a review cycle so your inventory, risk register and policy keep pace with new tools and new rules.
05Frameworks
How this maps to the frameworks
Useful for your auditor, board or procurement team. The mapping shows where this service supports each framework. It doesn't replace a formal assessment.
AI Security & Governance: framework mapping
Control
ASD Essentials
ISO/IEC 27001
NIST CSF
ISO/IEC 42001
AI policy and acceptable use
Governance and cultureExecutive artificial intelligence accountability
A.5.1, A.5.10
Govern (GV.PO-01)
A.2.2, A.9.2
Shadow AI discovery and AI inventory
Governance and culture
A.5.9
Identify (ID.AM-02)
–Not applicable
AI risk and impact assessment
Executive artificial intelligence accountability
–Not applicable
Identify (ID.RA-05)
A.5.2, A.5.3, A.5.4
Third-party AI vendor review
–Not applicable
A.5.19, A.5.23
Govern (GV.SC-07)
A.10.3
LLM and AI agent security testing
–Not applicable
A.8.8, A.8.29
Identify (ID.RA-01)
A.6.2.4
06Pricing
Pricing and how to start
Pricing*
Scoped after a strategy call
What changes the price
Price depends on your number of staff and business units using AI, how many AI tools and vendors need review, whether you need hands-on testing of AI systems you run and whether you are aiming for ISO/IEC 42001 certification.
Shadow AI discovery and an AI inventory
An AI acceptable use policy written for your organisation
No. ISO/IEC 42001:2023 is a voluntary international standard for an AI management system. Organisations adopt it to show customers, regulators or a board that their AI use is governed in a structured way. Some buyers, especially in government and regulated sectors, may start asking about it in procurement, so it is worth knowing where you stand even if you don't plan to certify.
What is shadow AI, and why does it matter?
Shadow AI is any AI tool staff use for work without the organisation approving it, such as a personal ChatGPT account or a browser plug-in. It matters because nobody has checked where the data goes, whether it is stored or used for training, or what the terms allow. Finding it is the first step, because you can't govern tools you don't know about.
Does the Privacy Act apply to our use of AI?
Yes, if your organisation is covered by the Privacy Act, which includes businesses and not-for-profits with annual turnover over $3 million, all health service providers and Australian Government agencies. For covered organisations, the Australian Privacy Principles apply whenever personal information goes into or comes out of an AI tool, including rules on use, disclosure, security and overseas transfers. Smaller businesses outside the Act still benefit from following the same practices.
What is prompt injection?
Prompt injection is an attack where someone crafts input that makes an AI system ignore its instructions. It can be direct, typed into a chatbot, or indirect, hidden in an email, web page or document the AI later reads. It is the first risk in the OWASP Top 10 for LLM Applications, and it matters most when an AI can take actions or reach sensitive data.
Is Catharsis certified to ISO/IEC 42001?
No. Catharsis is certified to ISO/IEC 27001:2022, the information security management standard. We help clients prepare for ISO/IEC 42001 and build their AI management system. Because both standards share the same management system structure, our experience running our own ISMS applies directly to the clauses the two standards have in common.
How is AI governance different from our existing security policies?
AI governance builds on your security policies rather than replacing them. Existing controls for access, data classification and suppliers still apply. AI adds new questions: whether outputs are accurate and fair, whether a tool learns from your data, who is accountable for automated decisions and how you test systems that respond to natural language. A good AI policy links back to your existing ones.
