Guidance
Which guidance covers data poisoning, and what it says
No single UK rule is about data poisoning. Three bodies of guidance cover it from different angles: the NCSC on secure development, NIST on attack types and OWASP on controls.
Last reviewed against Anthropic, UK AI Security Institute and arXiv. Sources
Use this page to pick the right document for the job. The sources describe each one as guidance, and none of them presents data poisoning as a legal duty.
The documents
What each one is for
NCSC guidelines for secure AI system development
Published on 27 November 2023, these follow a system through four stages of its life: design, development, deployment, and operation and maintenance [6]. Supply chain security sits under development, while deployment covers keeping the infrastructure and the models themselves safe from compromise [6].
NCSC machine learning principles
Version 2.0 was published on 22 May 2024. Each principle explains its goals, why it matters and how it could be implemented, and the principles underpin the NCSC guidelines [7].
NIST AI 100-2 E2025
OWASP LLM04:2025
The most practical of the three on poisoning. It sets out the affected stages, example scenarios and a list of mitigations covering provenance, versioning, vetting and testing [4].
Which to use
By role
-
1
Boards and risk leads: the NCSC guidelines, for the shape of the lifecycle [6]. -
2
Security architects: NIST, for shared terms when writing requirements [5]. -
3
Engineers and data teams: OWASP, for controls [4].
The three fit together. The NCSC gives the structure, NIST gives the vocabulary and OWASP gives the checklist, so a policy can borrow from each without contradicting the others.
A gap to note
The page for the NCSC guidelines, as read when this site was prepared, did not mention data poisoning by name [6]. Do not assume the guidelines are a poisoning checklist. Use them for lifecycle structure and read OWASP for specifics.
Where next
Turn guidance into action
See the prevention guide for the controls and the detection guide for testing.
Keep reading
Read next
What to do
A practical order of work for UK organisations to reduce data poisoning risk in AI systems: inventory, provenance, suppliers and testing.
Read the guide
Questions
Sourced answers on data poisoning in AI: what it is, how many documents it takes, whether it can be detected and what to do first.
Read the guide
Risks
The specific risks data poisoning creates for organisations that build, fine-tune or buy AI, with the evidence behind each one.
Read the guideHow CyPro can help
Need a hand planning for this?
Understanding the change is one job. Working out what it means for your systems, your suppliers and your roadmap is several more. CyPro's consultants can take any of them on.
Fractional CISO for the board
A senior security lead who advises the board on what to fix first, what to accept and what to tell customers, and owns the decision record.
More on cypro.co.uk
Hardening and segmentation
Our consultants put the controls in place that limit the damage when something is missed: firewall rules, network isolation and tighter admin access.
More on cypro.co.uk
Penetration testing
A focused test of your systems and the controls around them, so you know they hold before an attacker finds out for you.
More on cypro.co.uk
Vulnerability scanning
Find every internet-facing system you run, including the forgotten ones, and see which are still on a vulnerable build.
See the scanning service
24/7 security monitoring
Our SOC analysts watch your systems day and night for the signs of an attack in progress: unexpected processes, new outbound connections, logins that do not fit.
More on cypro.co.uk
Incident response
If you may already be compromised, our responders contain it, preserve the evidence, find out what was touched and get you running again safely.
More on cypro.co.uk
Plan ahead
Want a plan for your organisation?
Defending against data poisoning starts with knowing which AI systems you run, where their data comes from and who owns the risk. CyPro can help set the governance, risk assessment and controls around that.