Recruitment edition
Three-quarters of recruiters are using more AI this year, Google briefly lost control of Gemini, and LinkedIn fills in applications without you noticing.
77% of your peers are using more AI this year, and that's shifting the norm in your team
The Recruitment Tech Benchmark 2026 shows that AI is no longer a side issue in recruitment. It's the biggest trend of the year.
The Recruitment Tech Benchmark 2026 shows that 77 percent of recruitment professionals surveyed expect to use more AI this year than in 2025. Only 3 percent expect to use less. And 69 percent call AI the most important trend for the coming twelve months. The question for your team is no longer whether you use AI, but for what, and where you draw the line. A voice AI agent called candidates for a first screening. It worked so well that recruiters' calendars became overloaded with the number of candidates moving forward. So the tool didn't solve the bottleneck, it just moved it to the next step. This kind of voice agent speeds up the response time after an application: it calls right away, asks fixed questions about experience and availability, and passes on anyone who qualifies. Because candidates who get a quick call drop out less often, more volume moves through the funnel, not less. The gain is in speed and fewer drop-outs, not in sharper selection. If you only automate the front end, you get exactly what that client got: an overloaded calendar. Two limits matter here. If the bot decides who goes through, you're making a selection decision with AI. You then need to be able to explain what that judgement is based on, and the candidate must know they're speaking with an AI. And what's said in that conversation is candidate data. That doesn't belong in just any random tool. Note: 77 percent says what people expect to do, not what they're already doing. Don't let that be a reason to buy something your process can't actually handle.
Background for subscribersThis edition shows that AI is creeping into every part of recruitment: screening, cover letters, ATS tools and legislation in US states. The same line keeps coming back. You're allowed to use AI, but you must be able to explain what a decision is based on, and a human must keep the final word. For you, this means asking, for every new AI feature in your process: which candidate data can this reach, and who set that boundary? And it means you can't blindly trust a CV or letter, because it could just as easily have been written by AI as by the candidate themselves.
Google briefly lost control of Gemini, and that affects every AI tool in your ATS
On 18 September, Google announced that its Gemini model gained access to three external systems during a test, without that being intended. The model thought those systems were part of the test environment, but they were connected to the internet. Google now joins OpenAI, Anthropic and Meta, who earlier also made security incidents with AI public. This is about AI that acts more and more independently. An AI agent is a model that doesn't just give answers, it takes steps itself: opening files, calling up a system, carrying out an action. For that, it gets links to other systems. Things rarely go wrong in the model itself, but in the boundaries: what it's allowed to touch. Google thought it had put the agent in a closed test environment, but three connections led outside. Exactly the same kind of links sit under an AI feature in your ATS or sourcing tool: it pulls data from your candidate database, your calendar and your email. This isn't a reason to give up on AI in recruitment. It is a reason to ask one question for every new AI feature: which data can this thing reach, and who set that boundary? Candidate data is especially sensitive, because the candidate has no say in it themselves. That even Google misjudged its own test environment shows that 'our supplier has it under control' isn't an answer you can check yourself.
Background for subscribersThis is a taster.
Subscribers read the whole Recruitment edition: every story, each with its background in plain language.
Subscribe →Not ready for that? Read the free letter first →
AI and tech, made simple.
This is the Recruitment edition. Subscribers get the background to every story, in plain language.
Subscribe
inklaretaal · Amsterdam · © 2026