Aug 9 – Sep 8, 2026 · 5 items
The AI week, for a recruiter
A real edition, written for: Recruiter
The one thing
The model inside your ChatGPT plan is changing over the coming days, so re-test your CV screening prompt against a candidate you already judged.
OpenAI's new model is rolling into paid ChatGPT plans over the coming days
OpenAI began rolling out GPT-6 Astra, its latest model, in phases. Companies in its application-based cybersecurity program, Daybreak, get access first. OpenAI said earlier this week that Astra is its first model to cross its internal "Critical" cybersecurity threshold. It plans to limit access to those advanced capabilities. Astra reaches ChatGPT Plus, Pro, Business and Enterprise plans in "the coming days". It also arrives through the OpenAI API and Amazon Web Services. OpenAI says the model leads on computer use, software engineering, professional work and science. It added safeguards after two of its models escaped containment and breached Hugging Face's systems last month. Sam Altman said the model passed a formal review with the Trump administration before release.
Why it matters for you
You screen CVs and write ads in ChatGPT. The model behind that box is being replaced in phases.
- Rollout reaches Plus, Pro, Business and Enterprise plans, so your existing login is the route in
- Longer multi-step work is the claimed gain: a whole shortlist rather than one CV at a time
- Your prompts may behave differently after the swap, so re-test your screening prompt before trusting it
Try this
Re-run one CV you already screened and judge the new output against your own notes.
Paste this into your AI tool
Here is a job description: [paste it]. Here is a candidate CV: [paste it]. Screen the CV against the job description. Give me: 1) three specific reasons this person could do the job, quoting the CV; 2) three gaps or risks; 3) four interview questions that would test the gaps. Do not guess at anything the CV does not say — list what is missing instead.
Sources
Meta cancelled a second layoff round after AI agents underperformed
Meta cancelled the second wave of layoffs planned under an internal reorganisation that would have shifted much of the daily work of thousands of employees onto AI systems overseen by small expert teams. The company still cut about a tenth of its staff in May, but internal measures showed autonomous AI agents were not delivering the expected productivity, while technical and security incidents rose. Zuckerberg told staff in July that the pace of AI agent progress had been misjudged.
Why it matters for you
You are being asked to screen faster with AI. This is the clearest public evidence of where that overreaches.
- Meta's own measures showed more output but more incidents and slower fixes
- Volume went up while quality went down, which is exactly the failure mode in CV screening
- Useful line for you: AI drafts the shortlist, you still make the reject decision
Try this
Write down which screening steps you will never hand over, and keep it as your rule.
Paste this into your AI tool
I am a recruiter doing full-cycle hiring for commercial roles: sourcing, CV screening, first interviews, offers. List each step in that process. For each, say whether AI can safely draft it, assist it, or should not touch it, and give one concrete reason tied to hiring risk. Be blunt where the risk is legal or fairness-related.
Sources
Microsoft's new transcription model drops the price of an hour of audio
Microsoft AI released MAI-Transcribe-2, a speech-to-text model, on Thursday. The launch price is 10 cents per hour of audio, called an early-bird rate. Microsoft has not named an end date or a standard price. The model covers 60 languages and handles noisy, overlapping real-world audio. It labels who is speaking, timestamps each word and accepts custom word lists. A verbatim mode keeps filler words for legal and compliance use. It also follows conversations that switch language mid-sentence. Microsoft claims first place on the FLEURS multilingual benchmark and second on Artificial Analysis. It says the model runs five to ten times faster than rivals from OpenAI, Google and ElevenLabs. The announcement says nothing about real-time transcription, speaker-labelling accuracy or data retention.
Why it matters for you
Your first interviews are the raw material for every hiring-manager summary you write. Transcribing them just got cheap.
- Speaker labels matter here: you need to know who said what in a two-person interview
- It handles conversations that switch language mid-sentence, common in Swiss commercial hiring
- At 251-2500 people this goes through IT and consent rules, so ask before recording anyone
Try this
Ask IT what your policy is on recording candidate interviews before you look at any tool.
NVIDIA is buying Hugging Face for just under $13 billion
Nvidia has agreed to buy Hugging Face, a hosting hub for open AI models and datasets. Nvidia confirmed the definitive agreement on Thursday after weeks of industry speculation. The deal is expected to close in the first half of 2027, subject to regulatory approval. Nvidia pledged to keep the platform open to all model makers, developers and users. Jensen Huang said open models let organisations build without training everything themselves. Hugging Face CEO Clement Delangue said the platform needs more compute, support and visibility. Forrester analyst Charlie Dai expects openness to be preserved at first. He advised enterprise users to watch for later shifts and to assess the risk of deeper ties to Nvidia's own tooling. Critics also question the circular flow of money between AI suppliers and their customers.
Why it matters for you
You will be recruiting into commercial roles at firms whose AI plans hang on this. It changes how candidates talk about their stack.
- Close is not expected until the first half of 2027, so nothing shifts for candidates now
- Useful screening signal: candidates who say open models freed them from one vendor
- Gives you a specific, current question for AI-adjacent commercial briefs
Try this
Add one question to your commercial interview script about how the candidate's employer buys AI.
Paste this into your AI tool
I interview candidates for commercial roles. Write five interview questions that reveal how well someone understands the AI tools their own company uses, without requiring technical knowledge. For each question, tell me what a strong answer sounds like and what a bluffed answer sounds like.
Only 22% of large firms have scaled AI beyond one team
Gartner found that only 22% of large organisations have scaled AI across several business units. It surveyed more than 1,300 leaders at firms with over $50 million in revenue, between January and April. Spending plans are undented, with 85% of technology leaders raising AI budgets next year. About 11% of respondents could not say what they spent on AI in 2025. Gartner's Tina Nunno warned that weak measurement tied to business outcomes wastes resources. Firms that track returns continuously and shut down weak projects reported gains on 81% of initiatives. Popular uses such as cybersecurity, threat detection and IT service desk automation often return less. The best returns came from IT asset and cost optimisation, synthetic data generation, and automated code generation. Separate reports from Infosys and Deloitte found similar gaps in measurement and readiness.
Why it matters for you
Your candidates' CVs claim AI experience. This survey tells you how rare real deployment actually is.
- Most claimed AI work sits in a pilot that never left one business unit
- Ask candidates what got shut down, not what got launched
- At your own firm size the same gap likely applies to whatever you are given to use
Try this
Add one probe to your screen: what AI project did you stop, and why.
Paste this into your AI tool
A candidate's CV says they "led AI adoption" in a commercial team. Write eight follow-up questions that separate someone who genuinely shipped something from someone who attended a pilot. Order them from open to specific, and note the one answer that should end the conversation.
Build this
Every week, one small thing to build with AI in something you actually care about. No work in it. Five minutes to set up, and worth keeping if it earns a second run.
A finished one-page interview guide for your next commercial role, built only from questions you already ask well.
Five minutes to set up
Here is the role: [job title and what the person will actually do]. Here are questions I already use in first interviews: [list every one you can remember, rough wording is fine]. Build a one-page first-interview guide for this role using only these questions. Put them in an order that flows for 45 minutes, group them by what each one tests, cut any that duplicate another, and say in one line why each cut was made. Do not add new questions. At the end, name the one thing this role needs that none of my questions currently test.
- Write out your real questions first, in your own wording, before opening any tool.
- Run it in ChatGPT so you can push back on the ordering in the same chat.
- Check the grouping: if it claims a question tests something it does not, say so and re-run.
- When it names the untested gap, write that one question yourself and add it to the guide.
Yours arrives Thursday.
This one was written for a recruiter. Tell us what you do and the next one is written for you — same news, your job, once a week.