日本語 
Salesforce CEO Marc Benioff has spoken about the need for the artificial intelligence industry to take responsibility for regulating its own technology. He warned that companies could be sued and face government action if they do not address AI risks. His remarks were made at Salesforce’s Dreamforce conference in San Francisco where AI agents and enterprise automation were the topics.
Benioff’s remarks come amid a debate about how fast AI should grow and how much responsibility should lie with technology companies. Many technology leaders are talking about whether current safety practicesre enough as AI systems become more autonomous.
For businesses the question is no longer just whether to use AI. Companies now must decide how AI systems should be governed once they are linked to business data and allowed to act.
From AI Models to Autonomous Enterprise Systems
Salesforce is also moving toward an agent-driven software model.
At Dreamforce 2026 Salesforce introduced AIforce. AIforce is an interface layer that lets AI agents access Salesforce data, workflows, business logic, permissions and security controls across AI interfaces. Salesforce says this approach moves away from fixed software interfaces and, toward dynamic AI-driven interactions.
The company has also added job‑specific agents to Agentforce. These agents are made for sales, customer service, commerce, IT, HR and back‑office work. Salesforce says the newer Agentforce design lets agents pursue goals over times work with other agents and operate within business rules and permissions.
This change makes governance more important. An AI system that only makes a draft has one kind of risk. An agent that can change records contact customers manage transactions. Start business processes has another kind of risk.
Why Self-Regulation Matters
Benioff’s argument reflects a growing concern within the technology sector that companies need stronger safeguards before AI systems are given greater autonomy.
AI governance involves more than model accuracy. Organizations need to consider access permissions, data protection, audit trails, human approvals, security testing and the ability to stop or modify an agent when its behavior does not match business requirements.
Salesforce says its AI systems operate within customer-defined business rules, permissions and security controls. Its Agent Script technology is also designed to combine AI reasoning with deterministic rules.
For enterprise technology buyers, these capabilities could become increasingly important when comparing AI platforms.
Implications for Japan’s Enterprise AI Market
The discussion is particularly relevant to Japan, where companies are expanding AI use across manufacturing, financial services, retail, healthcare and customer operations.
Japanese enterprises often operate large technology environments containing sensitive customer, employee and operational data. Connecting AI agents to those systems could create productivity opportunities, but it also introduces questions about who can access information and what actions an autonomous system is permitted to take.
This could increase demand for AI governance platforms, identity and access management, cybersecurity monitoring, model evaluation and agent observability.
Japan’s technology companies could also find opportunities in developing industry-specific AI controls. A financial institution, manufacturer and hospital may require very different rules for how autonomous systems access data and execute tasks.
AI Regulation Could Become a Business Requirement
Benioff’s comments also highlight a potential shift in the economics of enterprise AI.
Organizations may increasingly expect vendors to demonstrate not only what an AI system can do, but how its actions can be controlled and audited. Features such as permission management, human approval workflows, traceability and data isolation could become part of procurement requirements.
This may create a growing market for AI assurance services and technologies alongside the AI applications themselves.
The regulatory question is also unlikely to disappear as AI capabilities advance. Governments are developing frameworks covering transparency, safety, data protection and accountability, while businesses are creating internal policies for responsible AI deployment.
A New Phase for Enterprise AI
The debate around AI regulation is occurring at the same time that companies such as Salesforce are pushing AI agents deeper into everyday business processes.
Also Read: Things Partners With AI Librarium on New Business
That combination is important. As AI moves from generating information to taking actions, governance becomes part of the technology architecture rather than simply a corporate policy issue.
For Japan’s enterprise technology market, this could mean that the next stage of AI adoption will be shaped not only by model performance, but also by security, accountability and control.
Salesforce‘s strategy demonstrates how quickly enterprise software is moving toward agent-based systems. Benioff’s warning adds another dimension to that transition: as AI becomes more capable of acting on behalf of organizations, the mechanisms for controlling those actions may become just as important as the intelligence behind them.
English