The comparison
Why prompt injection is not like SQL injection
SQL injection has a known cure. The NCSC argues prompt injection has no equivalent, and that treating the two alike leads teams to the wrong defences.
Last reviewed against NCSC, OWASP GenAI Security Project and Microsoft Security Response Center. Sources
The argument
What the NCSC wrote
On 8 December 2025 the NCSC published a blog post by Dave Chismon, its CTO for Architecture, titled "Prompt injection is not SQL injection (it may be worse)" [1]. Its central claim is that LLMs today keep no enforced separation between what they are told to do and what they are given to read [1]. By contrast, it says SQL injection is cured by parameterised queries [1].
The gap
Why the cure does not transfer
Parameterised queries keep data and commands apart. The NCSC says that for an LLM "there is only ever 'next token'", and that this attack may never be closed off as completely as SQL injection can be [1].
Research under way includes detecting attempts, training models to prioritise instructions and marking what is data, and the NCSC notes that all try to impose a distinction the technology does not natively make [1].
A better picture
The confusable deputy
The NCSC suggests thinking of an LLM as an "inherently confusable deputy" [1]. A classic confused deputy flaw can be fixed, whereas an LLM cannot be fully protected, so the goal becomes reducing risk and impact [1].
What it means for a decision
If a system cannot tolerate the residual risk, the NCSC says an LLM may be a poor fit for it [1].
Consequences
Four moves it points to
Read the controls in more detail on the prevention page, or see whether the problem can ever be fixed.
Keep reading
Read next
Prevention
Prompt injection controls for UK teams: least privilege, human approval, separating untrusted content and logging, from OWASP, Microsoft and the NCSC.
Read the guide
EchoLeak case
EchoLeak, CVE-2025-32711, was a zero-click prompt injection flaw in Microsoft 365 Copilot. What happened, what Microsoft did and what it teaches.
Read the guide
What is it?
What prompt injection means, in plain English: how text fed to an LLM can override its instructions, and why UK organisations should care.
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?
Prompt injection only shows up when someone tries it against your own tools. Penetration testing can probe the AI features you run or build, to see what a hostile instruction could reach.