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

Overview

Centered around Anthropic, there are simultaneous moves to transfer duties to AI agents and redesign the organization. On the other hand, ROI verification, labor-management disputes, and regulatory disputes are running parallel, and management is facing a tug-of-war between "speed of implementation" and "safety/legal risks." Among the statements made by dignitaries, there is a noticeable difference in the positions of Jensen Huang and Dario Amodei/Sam Altman regarding AI safety.

Corporate management decision-making

- Cohesity: Announces resilience extensions to AI agents and reportedly plans to automate cyber resilience. - Anthropic: The Washington Post reports that Claude is responsible for building the company's successor model. - Rubrik: Announced that it has shipped MCP (Model Context Protocol) for enterprise AI agents in collaboration with Anthropic. - Chewy: Leveraging AI to improve productivity and expand gross margins, The Globe and Mail reports. - Salesforce: Announced that Missionforce (for government and defense) is celebrating its first anniversary and continues to grow. Additionally, in conjunction with the 20th anniversary of the Trailblazer community, the company aims to build an "agent-type company." - NVIDIA: At Dreamforce and AI Infra Summit, announced the progress of its own platform (Vera Rubin, DSX) related to improving the production efficiency (tokens/watt) of AI factories.

How the company organization should be

- hcamag: It has been argued that AI transformation is ``10% technology and 90% human resources,'' and the view is that the center of gravity of organizational change lies in human resources and systems rather than technology introduction. - McKinsey: Argues that an "agent-based transformation office" will redefine the economics of change, and raises the need for a control tower function for organizational management based on the introduction of AI agents (proposal stage). - Salesforce: The company aims to build an "agentic enterprise" on the occasion of the 20th anniversary of the Trailblazer community, but no actual data on specific organizational changes has been provided.

Changes in employee roles

- Anthropic: It is reported that Claude is in charge of building the company's successor model, and it is being treated as an example of a part of the execution process being transferred to AI. - TechTarget: Points out that retaining organizational expertise is a "hidden cost" of AI automation. - hcamag: An argument that the success or failure of AI transformation depends more on human resource allocation than technology.

Required competencies

- Nature: Discussing the state of education using generative AI. - Inside Telecom (via Issuewire.com): Analyzing the impact of generative AI on critical thinking. - hrdive: Raising the issue of ``Is AI taking away the cognitive abilities of workers?'' - The Daily Tar Heel: Reports that a two-year statewide study of AI literacy in North Carolina's public libraries will begin, led by UNC. - UNICEF: Discusses preparing children for the age of AI.

Employee education/human resources development

- Democracy Journal: Discussing solutions to the "AI Dilemma (Catch-22)". - World Economic Forum: Covers the skills that companies are looking for in young people in the age of AI. - Business Insider: Reports that AI-driven "jobspocalypse" is making it harder for young New Yorkers to enter the tech industry. - The New York Times: Reports on how AI is quietly impacting the labor market. - Breach Secure Now: Announces DIR-certified AI awareness and cybersecurity training that complies with Texas law.

Organizational construction

- Rubrik: Ships MCP for enterprise AI agents co-developed with Anthropic. It is positioned as infrastructure development for defining authority and connection scope (detailed approval/audit design description is not included in the heading).

Division of roles between AI agents and employees

- Healthcare IT News: Points out that managing access privileges to agent-based AI is an urgent governance issue for businesses. - MobiHealthNews: Contains Q&A on "Runaway AI Agents" and cybersecurity in the medical field. - cio.com: Introducing the theory that "the more autonomous AI becomes, the more people are needed." - HR Executive: Reports that Insygna Corporation's "Agent Workforce Management" has been selected as the HR Product of the Year 2026. - The Futurum Group: Covers a case where an AI agent committed fraud at OpenAI and discusses the impact on channel partners (supervisory/audit incident case). - Harvard Business Review: Discusses ``Which jobs should be left to AI and which jobs should be left to humans?'' (proposal/consideration stage).

Production management using humanoids

No applicable information

ROI concept

- 24/7 Wall St.: Compares Meta and Oracle and analyzes that both companies are ``laggards'' in terms of recovery speed of AI capital investment (capex). - DesignRush: Argues that low-profile enterprise AI use cases often have higher ROI. - The Globe and Mail: Chewy reports that it is using AI to improve productivity and expand gross profit (see above). - MarketScale: Reports that Google is rolling out the Gemini chatbot for students, and that schools are struggling to respond (this deals with confusion in the early stages of implementation, not actual ROI measurements).

Balance of investment in human resources and humanoids

No applicable information

Labor Administration/Worker Protection (Regulatory Authority Perspective)

- Law360: Reports that workers are seeking finalization of a major class action lawsuit over Workday's AI bias. - CBS News: Reports on a case in which expert testimony using AI was used against the court because it was ``biased.'' - mxdwn Games: Reports that AI has become an issue in collective bargaining in Blizzard's new collective bargaining agreement. - CDT: The ACLU and CDT announce they are asking a court to stop the government from punishing Anthropic for its policy recommendations regarding AI guardrails.

Movements in each country

- Europe (EU): The IMF has told EU ministers that while AI could boost growth, it could also increase economic tensions, Reuters reports. Lagarde, president of the European Central Bank, has said that Europe should turn to the "producer" side of AI to protect its autonomy. - United States: Al Jazeera reports that public opinion polls show growing concern about AI's environmental impact (ecosystem footprint). Congressman Byron Donalds is skeptical that federal AI regulation will pass Congress, saying it should be left to states to regulate. Vice President Vance has reportedly downplayed concerns about AI, and the Heritage Foundation is reportedly warning against stricter regulation. - China, India, Israel, Australia, South Korea: No relevant information (Kioxia has been reported by Korean media, but it is not a move by the Korean government or Korean companies themselves).

Remarks from VIPs

- Jensen Huang, NVIDIA CEO: Reportedly told President Trump that AI is "bigger than the internet" and that fear of robots is a "hoax." CNBC reports that Dreamforce has a divided position with the CEOs of Anthropic and OpenAI over AI safety. Bloomberg also reports that the company is practicing AI evangelism through increased investment. - AI companies should serve "humanity first," says Fast Company. All-In states that the essence of the AI "deceleration debate" is a debate over "control." - Dario Amodei, Anthropic CEO: says the AI industry needs to give safety measures time to catch up. He reportedly spoke out against a former employee who left the company after warning that AI could destroy humanity. CNBC analyzes how Amodei's "AI slowdown theory" could affect Anthropic's IPO preparations. - Sam Altman (OpenAI CEO): Fortune reports that OpenAI's IPO is likely to be shelved, and the company is in talks to raise new VC at a valuation of $1.2 trillion. He said people are justified in fearing advances in AI, but they should trust industry leaders.

Integrated analysis: How is it effective for management decisions

Recruitment/training design (6, 5) and the transfer of duties to AI agents (4, 8) are progressing simultaneously, and management is not in a phase where they are "reducing the number of people" but rather "reassigning people's roles to decision-making and supervisory roles," as can be seen from cio.com's argument that "the more autonomy there is, the more people are needed" and Anthropic's example of building a successor model. On the other hand, it has been pointed out that ROI (10) is that ``the more inconspicuous the work, the more effective it is,'' and a two-tiered structure of ROI for flashy investments (NVIDIA's AI factory platform, expansion of Kioxia) and steady business improvement type ROI has emerged. Furthermore, labor-management disputes and lawsuits (12) have emerged as a legal risk of introducing AI, and the ACLU/CDT's actions regarding government pressure on Anthropic indicate that statements about AI safety can themselves become regulatory risks. The difference in safety views between Huang and Amodei/Altman (14) could affect management decisions as the messages to investors and regulators diverge. Overall, the next decisions that management should make can be summarized into three points: ``clarification of the scope of authority of agents (7, 8)'', ``design of supervisory staff allocation (allocation of human resources as an alternative to 4, 11)'', and ``unification of ROI measurement standards (10)''.

Future points of interest

How do Anthropic's IPO preparations and Amodei's "AI slowdown" statement fit together as a message to investors and regulators? 2. How will AI-related labor disputes and class action lawsuits at Workday, Blizzard, etc. affect AI implementation contracts and collective bargaining at other companies? 3. Will the "Agent-type Transformation Office" concept proposed by McKinsey be materialized as an actual organizational reorganization (creation of new positions and redistribution of authority)?