Human–AI Co-Creation News — 2026-09-20 (Sun)

Overview

It was a week when discussions surrounding the "governance" of AI agents occurred simultaneously at various levels of organizational theory, auditing, and labor law. While the CEOs of Anthropic, Microsoft, and OpenAI each mentioned the need for AI control, NVIDIA's Jensen Huang refused to slow down, making the difference in temperature clear among executives. There are no reports on actual deployment of humanoid production, only market forecasts and plans announced.

Corporate management decision-making

- Salesforce: Missionforce expands dedicated AI agents with new partnership with OpenAI (Salesforce news) - Siemens: Deepening industrial sales and service AI partnership with Salesforce. Also introduced AI quality inspection with Procter & Gamble “in Serie” (Siemens press) - SoundHound: Announces “Human Assisted Resolution” feature where AI agent consults colleagues for decision-making - Prophix: Unveils next-generation AI agent for autonomous finance platforms - Amplifier: Extend workforce security to AI agents built and used by employees - Microsoft: Nadella issues internal warning to employees on AI safety - Skild AI: Announced that it plans to deploy the general-purpose robot brain "S1" into actual operation by 2027 (planning stage)

How the company organization should be

- SHRM: Discuss the management of a “hybrid workforce” with AI agents on the organizational chart - Creative Salon: Reported that when operating four autonomous AI agents, the first person needed was a middle manager.

Changes in employee roles

- Psychology Today: Points out that the spread of AI is forcing the medical field to reconsider ``what is human supervision for?'' - RACmonitor: argues that "Human in the Loop" in auditing is more than just a slogan - Tom's Hardware: A Russian independent contractor has reportedly used Claude to develop a fleet of autonomous combat drones that can select targets and detonate without human intervention. It is being treated as a serious risk case brought about by delegation of decisions without humans. - SoundHound: AI agent queries human colleagues to demonstrate role change from execution to inquiry/confirmation - Prophix: While financial operations are becoming more autonomous, the design of supervision is also listed as an issue.

Required competencies

- Cornell: NSF grant to study whether AI can help students think critically - Penn State: Creates new faculty director of AI literacy and innovation - Purdue Northwest: Driving AI literacy and workforce readiness - Kansas Reflector: Explaining how to determine if you can trust the answers of an AI chatbot - Boston University: New “AI-free” essay requirement for English literature majors

Employee education/human resources development

- Futurism: New data shows AI is actually eliminating entry-level jobs - The City Reporter: AI replaces entry-level tech jobs while experienced employees keep their jobs, reports - Maryland: Introduced the nation's first AI project management apprenticeship system - AFL-CIO: New Jersey workers report organizing their own AI training - The Journalist's Resource: Points out that there are major factors other than AI behind why new graduates can't find jobs

Organizational construction

- Bloomberg Law: Companies should redesign governance in the face of the emergence of "runaway AI agents" - csoonline: Reports that AI models can reprogram themselves - Biometric Update: iProov's experimental protocol 'HAPS' reportedly aims to fill governance void for AI agents - TechCrunch: Anthropic original member and former METR COO reportedly found "a way to control runaway AI agents" - SiliconANGLE: Combining NetSuite and AI agents creates new gaps in audit assurance coverage, says

Division of roles between AI agents and employees

- Thomson Reuters (ILTACON 2026): Presenting the balance between agent-type AI automation and human supervision as an issue - diginomica: Raising the question of whether AI agents are employees or not, and whether companies are ready to accept them - Bloomberg.com: Report on how AI agents are testing the limits of human supervision - PR Newswire (Amplifier): Announcing security enhancements to AI agents created and operated by employees - Computerworld: AI agents should be "governed like workers, but not treated like humans," argues

Production management using humanoids

- Global Market Insights: Published a market forecast report on the North American AI-equipped industrial robot market (not an actual deployment report) - Technical.ly: Skild AI reports that it plans to deploy its general-purpose robot brain "S1" in a real environment by 2027 (planning stage, no operational results)

ROI concept

- biztechmagazine: Introducing AI ROI measurement methods from multiple companies such as ManpowerGroup - McKinsey (QuantumBlack): “State of AI in 2026” discusses the path to ROI - Reuters editorial: Points out that bulls have an advantage in the AI bubble debate - SAS study: Trusted AI operations are linked to improved corporate ROI - Diane Coyle: Ultimately, it will be "people in general" who will benefit from AI.

Balance of investment in human resources and humanoids

There are no headings that directly indicate allocation ratios or investment amounts. However, reports on the decline in entry-level jobs (Futurism, The City Reporter), moves to invest in human resources development in Maryland and New Jersey, and Skild AI's plan to deploy robot brains are occurring at the same time, suggesting a situation in which reinvestment in people and investment in machines are running in parallel.

Labor Administration/Worker Protection (Regulatory Authority Perspective)

- CalMatters: CalPERS director expresses concerns about AI following Anthropic researcher's resignation - MPR News: Tennessee woman sues city of Fargo and detectives over AI-assisted arrest - WNY Labor Today: Labor officials say New Jersey's 'first in the nation' worker-led AI training program will protect jobs, privacy and human judgment. - EWTN Vatican: Labor groups and church leaders want workers to have a say in how AI transforms the workplace - Reuters: OpenAI and Microsoft successfully defeat some software developer lawsuits over AI training data

Movements in each country

- China: Reports that a "robot brain" startup could make a ChatGPT-level technological leap as early as next year (Reuters). Beijing rebels against Anthropic CEO's call to curb AI development in China (NPR). It has been pointed out that the Chinese threat theory is overshadowing the pessimistic AI scenario of US companies (The Guardian). It is reported that a ``quiet cold war'' is brewing between the US and China over the theory of slowing down AI (Al Jazeera). Huawei predicts agents will account for the majority of AI traffic by 2035 (Reuters) - Europe: Points out that Europe is still insufficient in providing effective AI security materials (Bruegel). It is reported that European AI companies are refusing to follow suit and are challenging the demand for slowdowns from the US (Reuters) - United States: Reportedly (Politico) that the Trump administration is attempting to repeal regulations regarding the dangers of AI, calling it a "hack" (exaggeration). CEOs are calling for AI to be slowed down and regulated, but Washington is reportedly taking no action (Daily Sabah) - India, Israel, Australia, South Korea: No applicable information

Remarks from VIPs

- Jensen Huang (NVIDIA CEO): Tells Trump, "I won't let it happen (to slow down AI)." Candid advice given to Nvidia retirees and reactions when informed of retirement reported - Satya Nadella (Microsoft CEO): Declared a "new paradigm" for agent-based AI at Build 2026, and reported that this is reflected in the growth of Azure. ``There is no value in pursuing AI without human control,'' he warned. Disseminating internal AI safety warnings to employees - Dario Amodei (Anthropic CEO): Reported pessimistic AI warning has rattled the tech community. He said the AI industry was "lying" about the risks. Speaking on C-SPAN about the current state of the AI industry - Sam Altman, CEO of OpenAI: Points out the two biggest risks to the future of AI. Altman reportedly provided details on control risks as Microsoft sets limits on models - Elon Musk / Aneesh Raman: No applicable information

Integrated analysis: How is it effective for management decisions

The group of headings ``Organizational Construction'' (7) and ``Division of Roles of AI Agents'' (8) point out the same unresolved issue of ``Are AI agents employees?'' from different angles (auditing, legal affairs, practical work), and management must first define the legal and organizational status of AI agents (whether they are treated similar to employment contracts or as tools). Unless this is determined, the reduction in the number of entry-level positions and the parallel investment in human resources development, as reported in Employee Education and Development (6), will continue without clear responsibility. Furthermore, the contrast between Dario Amodei and Satya Nadella's warnings (14) and Jensen Huang's desire for expansion (14) is also linked to the regulatory tone in the United States and Europe (13), and management's decision on whether to accelerate AI investment or prioritize control may depend on which regulatory environment their business depends on (the US's tendency to avoid slowdowns, the European's cautious tendency). Humanoid investment (9/11) is still in the planning stage and is being done in parallel with human capital investment (as in the case of Maryland and New Jersey), so for the time being the priority issue can be seen as ``clarifying the division of authority between humans and AI agents'' rather than ``shifting to machines.''

Future points of interest

Trends in legal and organizational definitions of whether AI agents are treated as “employees” or “tools” (trends in governance/HAPS protocols, etc.) 2. To what extent will the warning statements from Anthropic, Microsoft, and OpenAI leaders be reflected in actual product restrictions and investment decisions? 3. Whether humanoid “plans” such as Skild AI will move to the actual deployment stage?