AI News Digest, July 17: HR’s Favorite AI Stat Has a Walkout Problem

Every HR trade publication ran the same headline this month. AI use in HR jumped from 26% to 43% in a single year. That’s a real number. Still, it’s worth celebrating on its own. But nobody put the second number in the press release. Because while adoption climbed, candidates started quietly rejecting the process itself. So there’s an AI interview transparency gap sitting underneath that growth chart, and it’s bigger than most HR teams realize. Below, that gap anchors a Friday roundup on AI spreading into Indian data centers, preventive health diagnostics, and a court that just got fooled by fake case law.

HR’s Favorite AI Adoption Stat Meets an AI Interview Transparency Gap

AI use in HR jumped from 26% to 43% in a single year. That’s according to SHRM’s State of AI in HR 2026 report. The survey polled 1,722 HR professionals in December 2025. So this isn’t gradual adoption. It’s a step-change. Still, most HR leaders are treating it as an unqualified win.

But here’s the number that didn’t make the celebratory posts. A separate 2026 survey of nearly 3,000 active job seekers found 63% have now been interviewed by AI. Of those, 38% walked away from a hiring process specifically because it included one, and another 12% say they would. The core issue isn’t the AI itself. Instead, it’s disclosure. Seventy percent of candidates were never clearly told upfront that AI would be evaluating them. In particular, one in five only found out once the interview had already started. (Source: 2026 Candidate AI Interview Report)

Where Undisclosed AI Interviews Stop Scaling

If you’re a 50-person startup running your first structured interview loop, one candidate quietly opting out barely registers. But if you’re screening 500 applicants a quarter for a Series B company, a 38% walkout rate is a hiring-velocity problem, not a footnote. Still, the fix candidates want isn’t complicated. First, 44% want upfront disclosure. Second, 39% want a plain explanation of what the AI is measuring. Finally, 46% want the option to request a human interview instead. None of that requires ripping out your AI tools. Instead, it just requires telling people they’re there.

What to do about it: Audit your ATS workflow this week. If a candidate can reach the AI-screening stage without clear, written disclosure beforehand, you already have the AI interview transparency gap. So fixing it is a policy change, not an engineering project.

Google Puts Gemini Inside India’s Borders, Airlock and All

Google announced at I/O Connect India 2026 that Gemini now runs entirely from data centers physically located inside India. The move is backed by a planned $15 billion infrastructure investment. So prompts, model weights and outputs never leave the country. Google calls the security layer a “physical airlock,” with supporting services fully disconnected from the public internet. (Source: Google)

This removes a trade-off Indian banks, hospitals and government agencies have lived with since generative AI arrived. Before, it was frontier models or data-residency compliance, not both. Now, they can have both at once. For any company running India-based teams in regulated sectors, this is the infrastructure argument for AI tools that HR and legal have been waiting for. Fewer sovereignty objections mean fewer procurement delays. In short, that timeline shift matters more than the headline itself.

Daniel Ek’s Body-Scanning Startup Raises $700M to Cross the Atlantic

Neko Health, the preventive-health startup founded by Spotify’s Daniel Ek, closed a $700 million Series C on July 15. Lightspeed Venture Partners led the round, while O.G. Venture Partners co-led. The round values the company near $7 billion, roughly four times its Series B mark from January 2025. (Source: TechCrunch)

Neko’s pitch is a 60-minute full-body scan paired with bloodwork and an AI-assisted consultation. The funding bankrolls a New York clinic as its US entry point. For HR leaders, this matters less as a health story and more as a preview. Specifically, preventive diagnostics like this are the exact category benefits teams will be fielding vendor pitches for within 18 months. So if you own benefits strategy, start asking what “AI health scan” coverage means for your plan design. Do it before a broker asks you first.

India’s Supreme Court Just Caught a Tribunal Citing Fake AI Case Law

India’s Supreme Court set aside orders from the National Company Law Tribunal on July 2. Specifically, the court found the tribunal had relied on AI-generated legal precedents that either didn’t exist or were fabricated. In Pooja Ramesh Singh v. Jammu & Kashmir Bank Ltd., Justices P.S. Narasimha and Alok Aradhe declared zero tolerance for unverified AI-generated authorities. In addition, they directed the Bar Council of India to draft guidelines. (Source: Verdictum)

This isn’t only a legal-industry problem. Because any HR team using AI to draft disciplinary memos, policy language or compliance documentation carries the same hallucination risk that just embarrassed a tribunal. So before you let an AI tool draft anything that could end up in front of a labor court, verify every citation by hand first. In short, that’s a five-minute check that prevents a five-figure legal bill later.

Quick Hits: Two More Signals Beyond the AI Interview Transparency Gap

• 29 countries just formed a China-led AI governance body. At the World AI Conference in Shanghai on July 16, 29 nations, including Russia, Brazil, Indonesia and Pakistan, signed an agreement establishing the World AI Cooperation Organization. The body is headquartered in Shanghai. Meanwhile, UN Secretary-General António Guterres attended the signing. (Source: Xinhua)
• A 27-billion-parameter model now fits on an iPhone. PrismML released Bonsai 27B on July 14. Its 1-bit form compresses to 3.9GB. Even so, it keeps over 90% of full-precision performance across 15 benchmarks, and it’s free under an Apache 2.0 license. (Source: 9to5Mac)

If today’s stories have you rethinking how AI shows up in your own hiring funnel, start with disclosure, not new software. For example, Asanify’s guide to AI-powered ATS platforms walks through what candidates should be told and when. In addition, if you’re building a team inside India, Asanify’s India EOR handles the compliance layer. That way, you can focus on the actual hiring decisions.

FAQ

What is the AI interview transparency gap?

It’s the mismatch between how fast companies are adopting AI in hiring and how clearly they’re telling candidates about it. SHRM found AI use in HR nearly doubled to 43% in 2026. Meanwhile, a separate survey found 70% of candidates were never told upfront that AI would evaluate them.

Do candidates actually leave hiring processes over AI interviews?

Yes. A 2026 survey of nearly 3,000 job seekers found 38% had walked away from a hiring process specifically because it required an AI interview. Still, another 12% said they would under the right circumstances.

Is AI-generated legal research reliable enough to use in HR compliance work?

Not without human verification. India’s Supreme Court set aside a tribunal’s rulings in July 2026 after finding they relied on fabricated AI-generated case law. So the same hallucination risk applies to any HR document, policy memo or disciplinary letter drafted with AI and not fact-checked before use.

Not to be considered as tax, legal, financial or HR advice. Regulations change over time so please consult a lawyer, accountant  or Labour Law  expert for specific guidance.