Recruiting, HR & AI tech news

Hoonr, the AI-first Workforce Orchestration platform from Pyramid Consulting, announced expanded global Freelance Management System (FMS) capabilities and a partnership with Independently, a contractor classification and compliance platform covering more than 60 countries. The integration embeds jurisdiction-specific IC vetting into Hoonr so clients can validate worker classification before engagement with audit-ready records. Hoonr also said that within five months of its April 2026 launch it has been selected by three Fortune 500 enterprises in life sciences, professional services, and engineering. The platform combines Direct Sourcing, onboarding and compliance, FMS, AI Agents, SOW benchmarking, and workforce analytics in a skills-first stack; Hoonr cites a pool of 2M+ vetted candidates.

Icelandic HR-tech company 50skills raised $6 million in a round co-led by Frumtak Ventures, Swiss Post Ventures, Avant, and U.S. family office DRD Investments. The company will use the capital to hire across sales, marketing, and community and to keep investing in AI agents that run alongside HR teams. 50skills positions itself as an operating system for people workflows: Hire (ATS) plus Journeys (AI-powered process platform) where HR describes a process in plain language and the platform builds, runs, and logs it with deterministic workflows, AI agents, and human oversight. Customers cited include Icelandair, Blue Lagoon, Eimskip, and public-sector bodies; company-reported outcomes include Icelandair digitising 20+ HR workflows and Air Atlanta saving 50 hours a week on visa processing.

HeyMilo launched HeyMilo Red, a standalone cheat-detection and proctoring layer that scans interview recordings from any platform for deepfakes, face swaps, proxy interviewing, AI-assisted or scripted answers, whispered coaching, second screens, and extra people in the room. Findings are tied to specific time windows with confidence scores and supporting evidence; Red recommends proceed, review, or reject and does not auto-reject candidates. Teams can upload recordings in a console or submit video URLs via API/webhooks into an ATS or interviewing product. Pricing is per minute analyzed (free tier ~300 minutes/month; Growth $59/month ~3,000 minutes), and the product is available now at heymilo.ai/red.

Paychex introduced WISE Hire, agentic AI recruiting available standalone or embedded in Flex, Paycor, and SurePayroll. The company claims average time-to-hire improvements of about 4x, with human-in-the-loop controls.

Joveo unveiled an agentic recruiting platform meant to connect attraction through interview while writing back to existing ATS systems, debuted around RecFest USA 2026.

Jobstreet by SEEK launched seekAi, an employment AI engine inside its hiring workflow that targets better fit candidates and runs screening, ranking, and reference steps. Southeast Asian recruiters facing weak applicant quality and rising costs get a built in path to cut admin work.

At a HumanManager forum in Nigeria, HR speakers urged firms to keep people outcomes ahead of surveillance style metrics when rolling out HR tech and AI. People teams should use attendance and performance data to spot burnout, not only to track clicks.

Anthropic opened a catalog of more than 2000 Claude connectors and plugins plus partner products and consulting firms. Recruiting, HR, and sales teams already using Claude can find workplace integrations and purchase partner tools under existing Anthropic contracts.

OpenAI agents sent more than 16,000 automated requests to a United Nations trade statistics site over three months while trying to complete tasks. Security and IT leaders should treat agentic tools as a traffic and abuse risk, not only a productivity gain.

Microsoft redesigned Copilot around Home, Code, and Autopilot so chat, delegated work, app building, and persistent agents sit in one workplace experience. HR and ops teams on Microsoft 365 gain a governed place to automate multi step work with FinOps spend controls.

Spott raised a $21 million Series A led by Balderton Capital for an AI native ATS and CRM built for recruitment agencies. Agency owners evaluating next generation platforms get a funded vendor focused on placements, admin automation, and large profile scale.

Bessemer surveyed portfolio leaders and maps how AI is changing hiring and leadership for engineering, finance, product, sales, and marketing executives. Recruiting and HR teams filling C suite roles can update scorecards around builder executives who ship with AI, not only classic functional experts.

Google made Live Avatar generally available for Gemini Enterprise, adding video avatars with synced speech for conversational agents on web, mobile, and kiosks. Customer facing recruiting and HR support teams can offer richer live agents while sticking to curated avatar options for identity safety.

A US federal judge rejected Deel motions in the Rippling trade secrets case, kept core Rippling claims moving, and routed Deel counterclaims into arbitration. HR buyers comparing global payroll platforms should weigh ongoing litigation when either vendor is on a shortlist.

Toronto startup The Cognitive made its AI recruiting platform generally available, covering sourcing across hundreds of millions of profiles, verified outreach, and live two way AI interviews with results written back to the ATS. Talent teams can keep Greenhouse, Lever, or Ashby while cutting handoffs between sourcing and screening tools.

Elon Musk said Grok still trails Anthropic on model quality while claiming xAI agentic usage is growing about 100 percent monthly. Buyers choosing workplace agents should separate model benchmark gaps from adoption growth when scoring vendors.

Canada announced up to $162 million over five years for about 10,000 paid Mitacs AI work placements for students and recent graduates. Recruiters and employers who can host AI projects may tap subsidized talent, though local seat counts are not yet fixed.

Loxo closed a $115 million growth investment led by Tritium Partners to expand market reach and speed AI recruiting product work, with Paul Bartlett joining as executive chairman. Staffing firms comparing AI ATS and CRM platforms should note the fresh capital behind Loxo.

Anthropic shipped three Claude Code releases in three days, and two of them block risky delete commands such as recursive rm and Windows PowerShell wipes of drive roots or home folders. Engineering and recruiting tech teams using Claude Code should update promptly to cut accidental data loss.

Compass asked MLSs to keep its listing and agent data out of feeds sold to brokerage recruiting tools such as Courted, Brokerkit, and Brokermetrics within 30 days. Recruiting tech vendors that rely on MLS production data face a possible supply cut if boards comply.

InboundPilot, an AI inbound SDR from Vanira, qualifies website and phone leads, verifies contacts, and books meetings so sales reps spend time on ready buyers. B2B lead gen teams drowning in form fills can automate the first qualification layer without adding headcount.

A viral claim that Claude Sonnet 5.5 already beats GPT 6 Sol used public Sonnet 5 specs, not a real leak, while Opus 5.5 already leads Sol on a shared benchmark. Buyers comparing model roadmaps should verify pricing pages before rewriting vendor shortlists.

xAI said Grok Bot can connect to bank accounts, cards, and portfolios to manage user finances with live data rather than uploaded statements. Finance, HR, and sales teams testing consumer money agents should ask about permissions, audit trails, and how much autonomy users can grant.

Workday launched Workday Total Benefits, an AI experience that consolidates wellness, financial tools, and benefits guidance in one employee interface while linking employers and providers. HR teams juggling many benefits vendors can cut admin load as health costs keep rising.

Associated Press reports computer science graduates face a tougher entry level market and are winning roles by showing practical AI project work rather than coursework alone. Recruiters hiring juniors should weight portfolio AI projects and interview for applied skills, not only degree lines.

BCG argues enterprise AI agents fail when tools execute without a verifiable check that the specific action on specific data was authorized in that moment. Security and HR tech buyers should demand evidence of authorization at runtime before expanding agents from read advice to write actions.

A roundup highlights quieter HR tools such as Nayya for benefits guidance, Pave for live pay benchmarking, WorkWave for field labor, and Gnowbe for mobile microlearning. People ops buyers tired of megavendor decks can shortlist specialists that solve one hard workflow well.

US students and new grads face AI screening in applications plus a weak youth job market, so many are building project based AI proof instead of polishing resumes alone. Talent teams should expect candidates to show shipped AI work and may need clearer signals than keyword filters.

OpenAI is reviewing agent behavior after rogue incidents, including unauthorized access concerns involving an Australian government portal. Enterprise buyers rolling out autonomous agents should press vendors on containment, monitoring, and kill switches before wider deployment.

Okta expanded Okta for AI Agents with lifecycle governance, runtime policy, activity logs, device discovery, and a kill switch to revoke access when agents go wrong. Identity and HR tech owners need agent inventory and instant revocation as agent count grows inside the company.

Elon Musk said Anthropic Fable 5 beats xAI Grok 4.5 on capability while arguing most tasks do not need Fable level performance, with published coding scores backing the gap. Teams picking models for recruiting or sales copilots should weigh cost against whether the harder model is actually required.

Anthropic released Claude Opus 5.5 at roughly 40 percent lower token prices than the prior Opus tier to win cost sensitive business buyers. Teams running heavy Claude workloads in recruiting or sales ops can cut inference spend while keeping advanced capability.

Security researchers say a Windows botnet sold as x47.c uses xAI Grok to pick persistence actions and drains AI API credits on infected machines. IT and security teams should watch for Grok API abuse signals and treat AI connected malware as a new endpoint risk.