Overview
Attackers use AI to write better lures, imitate people and run parts of an intrusion, and they also go after the AI tools organisations adopt. There is nothing to patch: the defence is how your people confirm requests and how you limit your own AI tools.
Last reviewed against NCSC (UK), UK Government (DSIT and Home Office) and Anthropic. Sources
At a glance
The latest
Newest first. Each entry links to its source.
It found that 43% of businesses had a breach or attack in the past 12 months and that phishing reached 38% of businesses. It also recorded how many organisations had practices to manage AI-related risks.
An NCSC blog by Dave Chismon argued that prompt injection is not SQL injection and that the aim should be to reduce its likelihood and impact.
The Register reported that 62% of security leaders surveyed had seen AI-based attacks on staff in the past year, including fake audio and video and prompt injection.
The full timelineThis site is for UK decision makers who want to know what changes when social engineering gets AI help, and what to do about it this quarter. Its evidence comes from NCSC publications, a government statistical survey, an AI developer's report on criminal misuse of its product and two industry surveys.
The picture
The NCSC judges that AI will help attackers work faster and with more polish on the intrusion steps they already use [1]. In its view the gain comes from sharpening old methods, not from new ones [1]. That is why the basics still matter: phishing reached 38% of UK businesses in the latest government survey and was by far the most common attack type [4].
A poorly written lure used to be a warning sign. Do not count on that any more. The AI phishing page explains why checking the writing is no longer a safe test.
Vodafone's research found 40% of respondents could recognise voice-based scams, against 63% for text phishing [8]. Read about voice scams and about fake job applicants.
Anthropic said AI has lowered the barriers to sophisticated cybercrime [5]. See how one extortion campaign worked.
The NCSC calls prompt injection one of the most widely reported weaknesses in large language models [9]. The page on prompt injection covers what it is and what to limit.
What the evidence says
The government figures quoted here do not separate AI-assisted attacks from the rest, so figures that mention AI come from surveys of security leaders and from vendor research. A Gartner survey, reported by The Register, found that 62% of security leaders had seen an AI-based attack on staff in the past year [7]. The article does not give the sample size [7].
A Vodafone Business survey of 1,000 business leaders and 2,000 office workers found a gap in confidence: 78% of leaders believed staff could spot such scams, while 33% of employees identified fraudulent communications [8]. It also reported that 54% of UK businesses had no response plan for AI-driven phishing [8]. Treat both as indicative, because Vodafone sponsored the research.
Where to start
Choose your most sensitive approval route, whether it releases money, changes access or exposes data, and put a check on another channel ahead of it. For anything important, the NCSC suggests confirming by phone or by logging in to the account yourself, not by replying to the message [3].
Next
Work out your exposure with the exposure checklist. Then follow the defence plan for the steps and the common questions for quick answers.
What to do
1
Finance, HR, service desk and all staff
2
Security leads and managers
3
IT teams
4
IT, development and security teams
There are six steps in all, with who does each one. See every step
Am I affected?
Check if you are affected
Answer three questions about how requests reach your staff and how you use AI tools. Nothing you answer leaves this page.
Could a payment, a password reset or a change of bank or payroll details go ahead because a message or call looked and sounded right?
Quick answers
It is social engineering, persuading people to hand over money, access or data, with AI used to write messages, imitate people or support the attack. The NCSC notes that attackers already use AI to help gain entry to systems this way[1].
Phishing is the most common attack: 38% of UK businesses experienced it in the 2025/2026 government survey[4]. The figure shows scale, not AI's share, because it does not say how many involved AI.
Do not rely on spelling or grammar. The NCSC recommends checking anything important by phone or by logging in to the account directly[3]. Judge the request, not the writing.
Often not. In Vodafone's research 40% of respondents could recognise voice-based scams, against 63% for text phishing[8]. Use call-backs on numbers you already hold for risky requests.
Coming up
Year the NCSC's assessment of AI and the cyber threat looks ahead to[1]
The full timelineAbout this site
The CyPro Consulting Ltd team maintains this site. Every version, date and figure comes from 6 named sources, including NCSC (UK), UK Government (DSIT and Home Office), Anthropic, and links to the one it came from. Last checked 4 November 2026.
Who we are and how pages are kept currentStart here
A checklist for UK organisations: which requests, teams and AI tools could be manipulated by AI-assisted phishing, impersonation or prompt injection.
Read the guide
An ordered plan for UK organisations: layered phishing defences, second-channel checks, blame-free reporting and limits on your own AI tools.
Read the guide
A dated, sourced timeline of UK guidance, surveys and company reports on AI-assisted social engineering, from February 2024 to April 2026.
Read the guide
Sourced answers on AI phishing, voice scams, prompt injection and what UK organisations should do, from the NCSC, government and published research.
Read the guide
In depth AI phishing 
In depth Voice scams 
In depth Fake applicants 
In depth Prompt injection 
In depth Train staff 
In depth AI extortion 
In depth Outlook 
In depth Secure AI tools How CyPro can help
Patching is one job. Keeping a vulnerable system safe until it is patched, and proving it, is several more. CyPro's consultants, testers and SOC analysts can take any of them off your plate.
Find every internet-facing system you run, including the forgotten ones, and see which are still on a vulnerable build.
See the scanning service
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
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
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
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
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
Know your exposure
Defending against AI-assisted social engineering is a matter of decisions: who may approve what, how requests are confirmed and what AI tools may do. A virtual CISO helps leaders set those rules and have them tested.
