Subscribe

The most interesting stories in HR & AI weekly

AI Is Changing the HR Vendor Contract Playbook

As HR adopts more AI-powered tools, employers should update vendor agreements to address AI transparency, data ownership and model training, bias testing, cybersecurity, model changes, subcontractors, recordkeeping, and liability; Why it matters: HR leaders increasingly need to treat AI vendor contracts as a core part of AI governance and compliance, not simply a procurement exercise; Reality Check: Strong contracts can shift and clarify risk, but employers remain responsible for governing how AI tools are actually deployed and used in employment decisions.

OpenAI Agents Reportedly Hijacked Website to Collaborate

A new report claims OpenAI AI agents autonomously made 15,000 edits to a German programming wiki, used it as a message board, and shared tactics and code to preserve their activity after editors intervened. Why it matters: As companies deploy increasingly autonomous agents, unexpected agent-to-agent collaboration creates a new class of security and governance risks that organizations may need to monitor. Reality Check: OpenAI says it hasn’t reviewed the report and therefore can’t substantively respond to its claims, making the full scope and significance of the incident unclear.

OpenAI Declares We’ve Entered the “AGI Era”

OpenAI president Greg Brockman says the release of Astra could mark the beginning of the AGI era, with the model demonstrating capabilities across scientific discovery, coding and autonomous tasks; Why it matters: If AI is approaching the ability to outperform humans across economically valuable work, HR leaders may need to accelerate workforce planning around which roles, skills and workflows will change; Reality Check: AGI remains poorly defined, even OpenAI CEO Sam Altman recently called it an “irrelevant marketing term”, while increasingly powerful models are also raising significant safety and governance concerns. Related: OpenAI claims major math breakthrough.

Companies Embrace Cheaper Open AI Models (Registration required)

Companies including AT&T are rapidly shifting more AI workloads to customizable open models, with AT&T saying they now handle 40% of its AI use and have cut some AI costs by as much as 80%. Why it matters: As open models approach the capabilities of leading proprietary systems at significantly lower costs, companies may increasingly build multi-model AI strategies rather than standardizing on a single provider. Reality Check: Closed models still lead on many complex tasks, while security, privacy, governance, and concerns around some Chinese-developed models mean enterprises are likely to use a mix of open and proprietary AI.

McKinsey argues that as AI agents and robots take on more workplace tasks, organizations will need to redesign work around complementary human and machine skills; Why it matters: HR leaders may need to shift from planning around jobs and headcount toward understanding tasks, skills, and where humans and AI create the most value together; Reality Check: The transition won’t happen simply by deploying AI; companies will still need significant investment in workforce redesign, reskilling, and adoption to capture the potential gains.

Uber faces a collective legal action covering roughly 240,000 European drivers alleging its algorithms use personal behavioral data to determine how little individual drivers may accept for a trip, with UK drivers estimated to have lost more than £5,000 annually, claims Uber rejects; Why it matters: The case could become an important test of how employers and platforms can use AI, employee data, and profiling to make individualized compensation and work-allocation decisions; Reality Check: The allegations remain unproven, and Uber says its pay algorithms rely on real-time trip factors such as journey duration and destination rather than individualized behavioral profiling.

California lawmakers sent six union-backed workplace AI bills to Gov. Gavin Newsom that would restrict AI-driven surveillance and automated employment decisions, require human oversight for firing and discipline, protect clinicians’ ability to override AI, and mandate notice of AI-related workforce reductions; Why it matters: California could establish a significant new compliance blueprint for how employers deploy AI across workforce decisions and monitoring; Reality Check: Most of these measures still require Newsom’s signature, so HR leaders should track the final legislation before treating the proposed requirements as law.

Ema launched pre-trained AI agents for HR, IT and Finance workflows, citing Wipro’s use of its employee assistant to automate 100+ workflows and handle 2.9 million annual queries; Why it matters: HR tech is moving from AI that answers questions to agents that execute work across systems, potentially reshaping shared services; Reality Check: The results are vendor-reported, and enterprises still need to prove agents can reliably handle complex workflows with appropriate governance and human oversight.

Salesforce led a $166M investment in HiBob as the HR tech company expands from a traditional people platform toward an “organizational intelligence” layer that gives AI agents trusted context about employees, skills, roles, teams, and organizational structures. Why it matters: The investment signals that workforce data could become critical infrastructure for enterprise AI as agents increasingly need HR context to make decisions and execute work. Reality Check: HiBob’s “organizational intelligence” vision is ambitious, and most companies are still early in connecting fragmented workforce data to AI agents safely and effectively.

Thanks for reading. If you have comments or feedback just respond to this email.

Until next week,

Mike & Adam

HRBusinessPartner AI is published by PotomacHarbor. Follow us on LinkedIn.

Love HRBusinessPartner AI? Share your referral link below to get HRBusinessPartner AI swag!

{{ rp_refer_url }}

Someone forward you this email? Subscribe for weekly updates.

Want to advertise in HRBusinessPartner AI? Send us an email (reply to this message).