Checkr Launches Deepfake Detection for Background Checks — Microsoft Teams Integrates It
Background check giant Checkr has added deepfake detection to its screening platform, alongside Microsoft’s announcement that Teams will integrate synthetic media detection for interview recordings starting July 2026.
Checkr’s tool analyses voice and video submissions for signs of AI manipulation — detecting synthetic voice, video face-swapping, and AI-generated written content. Microsoft’s Teams integration flags whether interview responses show signs of AI generation in real-time or in recorded playback.
The tools are aimed at the growing problem of “deepfake job interviews” — candidates using AI to generate responses, change their appearance, or alter their voice during remote interviews. A survey cited by Checkr found 34% of recruiters reported suspecting AI-use in candidate communications.
Why it matters: The arms race is here. Candidates use AI to optimise applications and interviews. Employers deploy AI to detect that AI use. Next comes AI that detects the anti-AI detection. The framing matters: Checkr and Microsoft position this as “integrity protection,” but critics see it as surveillance creep — candidates should be assessed on outcomes, not on whether they used the same tools the employer is deploying.
Hermes Germany Ends AI-Powered Hiring Pilot After Accuracy Concerns
Hermes Germany, the logistics and parcel delivery company, quietly ended its AI-powered hiring pilot after workers raised concerns about inaccurate candidate assessments. The system, which used computer vision and NLP to evaluate video interviews, reportedly flagged candidates as unsuitable based on irrelevant factors — accent differences, background lighting, or nervous fidgeting.
The pilot ran for eight months across 12 locations. Worker councils challenged the system’s validity, and independent testing revealed accuracy rates below 70% for key candidate attributes. Hermes has reverted to human-led recruitment while it evaluates alternative approaches.
Why it matters: This is the reality behind the AI hiring hype. The tools are still unreliable for nuanced assessments, especially in roles requiring social or physical skills (delivery drivers, warehouse workers). Every failed pilot makes it harder for the industry to argue that AI hiring is ready for prime time. Expect regulators — especially in the EU under the AI Act’s high-risk classification — to cite this case in rulemaking.
AI Fluency Is Becoming the New “Proficient in Microsoft Office”
A wave of job postings across professional services, marketing, and operations roles now include “AI fluency required” as a baseline qualification — not a bonus. Inc Magazine reports the phrase appears in 22% of all new US professional job postings, up from 4% in early 2024.
What constitutes “AI fluency” varies wildly. At the low end: knowing how to prompt ChatGPT or Claude effectively. At the high end: building AI agent workflows, evaluating model outputs, and understanding AI limitations.
Fast Company notes the trend is creating a new class divide in hiring: professionals who can articulate how AI transforms their domain get interviews and offers; those who can’t are filtered out before the first round. The gap widens by age — younger workers with more AI exposure have a measurable advantage.
Why it matters: “AI fluency” is following the trajectory of “computer literacy” in the 1990s and “social media marketing” in the 2010s — a differentiating skill that becomes a baseline requirement within 2-3 years. The difference is the speed: this transition is happening in months, not years. If you can’t demonstrate AI competency in your field, you’re already behind.
Intelligent.com: The 10 Best AI Degrees for 2026
Intelligent.com released its 2026 ranking of the best AI degrees, evaluating over 200 programs across enrolments, cost, faculty expertise, and placement outcomes. The top 10:
- Carnegie Mellon — BS in AI (first dedicated AI undergrad in the US)
- MIT — BS in Computation and Cognition
- Stanford — BS in Symbolic Systems (AI track)
- UC Berkeley — BS in Data Science (AI/ML concentration)
- Georgia Tech — BS in Computer Science (AI/ML thread)
- University of Washington — BS in Computer Science (NLP/AI focus)
- University of Toronto — BSc in AI/Machine Learning
- ETH Zurich — BSc in Data Science
- University of Edinburgh — BSc in AI and Computer Science
- Purdue — BS in AI
Notably, four-year AI degree programs are still relatively rare — most schools offer AI as a concentration within CS. Carnegie Mellon’s dedicated AI undergrad remains the gold standard.
Why it matters: The ranking reveals a structural education gap. Only a handful of universities offer dedicated AI degrees. Most students are in CS programs with AI concentrations — which may not teach the agentic AI skills the market actually demands. If you’re choosing a program, look for courses on: agent architectures, AI safety, human-AI interaction, and evaluation/red-teaming — those are the 2027 skills.
🔍 THE BOTTOM LINE
Three career dynamics are colliding: AI fluency baseline requirements are spreading faster than any training infrastructure can support; AI hiring tools are failing their first real-world tests (Hermes pilot cancellation); and employers are arming up against AI-assisted candidates with deepfake detection in interviews. The message to professionals is clear: learn AI use, but be prepared to prove it’s genuinely your work.
❓ Frequently Asked Questions
Q: Should I be worried about deepfake detection in interviews? If you’re using AI to generate answers you couldn’t produce yourself — yes. But most professionals use AI for preparation, research, and structure, which is distinct from real-time response generation. The detection tools currently available have high false-positive rates.
Q: What happened at Hermes Germany and what does it mean? Hermes ran an AI interview analysis pilot for eight months. It was inaccurate (<70%) and flagged irrelevant factors like accent and background. Worker councils challenged it, and Hermes ended the pilot. This is a real-world case that regulators will use to set standards for high-risk AI hiring tools under the EU AI Act.
Q: What’s the best way to build AI fluency? Domain-specific practice. Learn how AI applies to your actual job — marketing professionals using AI for campaign analysis, accountants for reconciliation, lawyers for contract review — not generic prompt engineering courses. The value is in the application, not the tool itself.
SOURCES
- Checkr Blog — Deepfake detection in background checks
- Microsoft — Teams synthetic media detection for interviews
- Windows Central — Microsoft Teams deepfake detection incoming
- Handelsblatt — Hermes Deutschland beendet KI-Recruiting-Pilot (German)
- Fast Company — AI fluency is the new baseline qualification
- Inc Magazine — AI fluency appears in 22% of new professional job postings
- Intelligent.com — Best AI degree programs 2026
- Yahoo Finance — Intelligent.com best AI degrees 2026