The talent pool is real but tight. There are approximately 143,000 individuals in the UK cyber security workforce, with growth moderately accelerating from 2% in 2022 to 5% in 2024 the GOV.UK Cyber security skills in the UK labour market 2025 report. That growth is encouraging, yet demand keeps running ahead of supply.
The pressure shows up at the point of hire. A comparison with the Cyber Security Skills 2024 report showed 70% of cyber firms reported at least one hard-to-fill vacancy the GOV.UK AI Labour Market Survey 2025. When seven in ten firms can't fill a role, specialist security skills become the bottleneck for everything else.
And the AI layer is new pressure on top. Demand for AI skills increased nearly 200 percent in a year, with London accounting for 80 percent of AI-related job postings and nearly two-thirds of all technology vacancies in the UK Accenture data reported by The Register. Put those two trends together, cyber scarcity and an AI skills surge, and you get a brand new specialism: people who can red team AI agents.
Autonomous agents don't behave like traditional software. They take actions, call tools, chain tasks and make decisions, so the ways they fail are different too. Red teaming an agent means thinking like an attacker: can you trick it with a crafted prompt, make it leak data, push it past its guardrails or get it to act on instructions hidden in the content it reads?
This sits at a crossroads. It blends offensive security skills, which are already scarce given that 70% of cyber firms reported at least one hard-to-fill vacancy the GOV.UK AI Labour Market Survey 2025, with AI skills that are suddenly in huge demand after rising nearly 200 percent in a year Accenture data reported by The Register. Few people hold both. That's exactly why it reads as a specialism rather than a line on a job description.
There's a geography angle too. With London accounting for 80 percent of AI-related job postings Accenture data reported by The Register, the obvious talent is concentrated, and employers outside the capital need to think harder about where they look and how they attract people.
Start with the offensive security foundation: penetration testing, adversarial thinking and a genuine curiosity about how things break. Layer on AI fluency: how large language models and agents work, prompt injection, data poisoning, jailbreaks and model evaluation. The best people can translate an abstract risk into a repeatable test and then write it up so a non-specialist can act on it.
Because the crossover skill set is rare, cast a wide net. Look at strong offensive security people who can be grown into AI red teaming, and at AI specialists with a security instinct. Given that 70% of cyber firms reported at least one hard-to-fill vacancy the GOV.UK AI Labour Market Survey 2025, waiting for the perfect ready-made hire can cost you months.
Test for the right things. Use a practical exercise: hand someone an agent and ask them to find three ways to make it misbehave. That tells you more than a CV. Be clear about clearance requirements early, since vetting can add real time in this sector. And move quickly once you find the right person, because, with London accounting for 80 percent of AI-related job postings Accenture data reported by The Register, strong applicants have options.
When the skills are rare and the clock is against you, speed and reach matter. We search a database of 15 million candidates to find the right person, and we manage the hire end to end, from a ranked shortlist through to interview. Our recruitment agent manages recruitment end to end for 8% on a successful hire, with no monthly fee and no upfront cost, through Reed.ai. Tell us what your red teaming role needs today, and we'll get started.
