The most quotable number in the AI jobs debate this week is a hiring one, not a layoff one. At LinkedIn’s Talent Connect conference on 29 September 2026, CEO Dan Shapero presented new LinkedIn research showing that AI safety roles grew 91 percent over the past year — even as he described overall tech hiring as largely “frozen.” Forward-deployed AI expert roles, the people who embed AI systems into client operations, are up 226 percent since 2023, and AI forensic analyst roles rose 169 percent.
🔍 The Bottom Line
The jobs market’s AI story is currently less “robots take your job” and more “entirely new job titles are being invented faster than the market can hire for them.” LinkedIn’s growth numbers are one platform’s lens and they measure roles that barely existed three years ago — but the direction of travel is unmistakable, and it is not toward a shrinking tech workforce.
The Numbers Behind the 91 Percent
LinkedIn’s research puts AI safety work — technical evaluation, risk mitigation, policy, and building safety frameworks into products — among the fastest-growing job categories in a stagnant market. Shapero’s framing at the conference: “AI safety is something a lot of people are talking about.” The Business Insider write-up of the research, published 30 September, also reports LinkedIn has counted roughly 2 million new jobs created by AI technology since 2023.
The salary ceiling in the field is substantial, though advertised pay at frontier labs should be read as an outlier layer, not a market average: the report cites OpenAI safety-team data scientists at up to $325,000 a year, Anthropic safeguards enforcement analysts at up to $285,000, Google kids-and-families policy strategists at up to $239,000 and Meta integrity program managers at up to $209,000.
Growth Titles Nobody Was Hiring For in 2023
Forward-deployed engineers — a role we named the standout AI job path earlier this year — have grown 226 percent since 2023 on LinkedIn’s numbers, alongside AI forensic analysts at 169 percent. Neither title featured meaningfully in conventional career advice three years ago. Both now sit inside ordinary corporate org charts: hospitals, banks and retailers are hiring people who can evaluate, audit and troubleshoot AI systems in the field.
That pattern fits the wider evidence. US Bureau of Labor Statistics projections running through 2035 forecast 21 percent growth for information security analysts — the adjacent, security-focused career ladder. And Lightcast data reported by CFO.com shows US job postings seeking AI skills rose from 9,241 in January 2022 to 80,597 in August 2026 — a 772 percent jump — with median advertised compensation up 39 percent to $147,500.
The Catch: Demand Is Rising Faster Than Pay
The Lightcast numbers carry a warning worth reading alongside LinkedIn’s optimism. Job postings grew 772 percent while pay rose 39 percent — real growth, but a fraction of demand’s pace. Payscale’s companion survey, cited in the same CFO.com report, found 49 percent of companies say their salary structures have not kept pace with AI, 40 percent cannot find the AI skills they need, and 58 percent either pay a premium for AI skills now or expect to within a year. Payscale’s sharpest line concerns the internal market: current employees who upskill watch new hires command 20 to 40 percent premiums they are not getting — “that’s a retention time bomb.”
For workers, the honest synthesis is that AI skills currently buy you a better chance at hire more reliably than a guaranteed pay leap — evidence we have tracked all year in both directions, including the broader wage premium data from Indeed’s listings and the counter-signal that AI exposure can come with wage pressure. Which of those forces dominates depends heavily on seniority and sector — the same lesson the senior-heavy pay research pointed to this week.
What It Means for New Zealand
New Zealand will not grow a frontier-lab safety sector, and the $325,000 salaries are not coming to Wellington. But the demand edge is already visible locally: SEEK’s New Zealand data this year has shown AI skills appearing in a rising share of job ads on this side of the Tasman, and security-adjacent work is one of the categories NZ employers consistently list as hard to fill. The transferable lesson from LinkedIn’s data is the one that matters for a small market: adjacent-title pivots — QA engineers into AI evaluators, security analysts into AI safety analysts, support leads into forward-deployment-adjacent roles — are where growth concentrates, and none of them require a frontier-lab postcode.
For a country with unemployment sitting at multi-year highs, “AI safety analyst” is not the job to bet a career on. But the pattern behind it — audit, evaluation, oversight and deployment roles growing faster than builder roles — is exactly the shape a small labour market can actually catch.
❓ FAQ
How fast are AI safety jobs growing? LinkedIn’s research, presented at Talent Connect on 29 September 2026, puts growth at 91 percent over the past year — against an overall tech hiring market its CEO describes as largely frozen.
What other new AI job categories are growing fastest? Forward-deployed AI expert roles are up 226 percent since 2023 and AI forensic analyst roles are up 169 percent, per LinkedIn’s research.
How much do AI safety jobs pay? Advertised pay at leading labs reaches $325,000 (OpenAI safety data scientists) and $285,000 (Anthropic safeguards analysts), but these are top-of-market figures; US-wide, median advertised compensation for AI-skilled postings is about $147,500 per Lightcast.
Is this relevant in New Zealand? Directly equivalent roles are rare, but the adjacent pattern — AI evaluation, security and deployment-support skills in rising demand — matches what NZ job-ad data has shown all year, and pivots from security, QA and support careers are the practical local pathway.
🔍 The Bottom Line
A frozen market is not an empty one. The fastest-growing AI job categories are oversight and deployment roles — safety, evaluation, forensics, forward deployment — and they grew out of almost nothing in three years. For anyone planning a move, the data says the opportunity is narrower than “learn AI” but realer than the hype: get adjacent to the systems companies have already bought, because the people who mind them are the ones being hired.
📰 Sources
- LinkedIn research presented at Talent Connect, 29 September 2026
- Business Insider, 30 September 2026
- Yahoo Finance, 30 September 2026
- US Bureau of Labor Statistics employment projections
- Lightcast data via CFO.com, 24 September 2026
— CJ Murden, editor of Singularity.Kiwi. Former digital technologies teacher, author of AI-focused books. Writing with a New Zealand focus.
