This article traces the history of calls for AI regulation, from early warnings by executives to current proposals for laws and standards, while examining why voices from industry leaders carry both influence, contradictions, and conflicts of interest.
The key point is that these calls are not only about safety. They also concern business direction, the power to set rules, and social responsibility.
This article traces the history of calls for AI regulation, from early warnings by executives to current proposals for laws and standards, while examining why voices from industry leaders carry both influence, contradictions, and conflicts of interest.
The key point is that these calls are not only about safety. They also concern business direction, the power to set rules, and social responsibility.
From Warnings on Stage to Policy-Level Issues
Calls for AI regulation began with warnings from executives on public stages before evolving into proposals concerning laws, standards, and corporate accountability. Technology is therefore no longer viewed merely as a product, but as something that affects society as a whole.
As the issue entered the policy arena, the influence of executives grew alongside questions about conflicts of interest, because the same people may both call for rules and have a stake in those rules.
From Warnings on Stage to Policy-Level Issues
Calls for AI regulation began with warnings from executives on public stages before evolving into proposals concerning laws, standards, and corporate accountability. Technology is therefore no longer viewed merely as a product, but as something that affects society as a whole.
As the issue entered the policy arena, the influence of executives grew alongside questions about conflicts of interest, because the same people may both call for rules and have a stake in those rules.
When AI Enters Our Lives but the Rules Cannot Keep Up
Imagine seeing a deepfake video and being unable to tell what is real, or applying for a job only to be rejected by an automated system without any explanation. These problems are not remote or abstract, because they affect the credibility, income, and opportunities of ordinary people.
When companies use AI to make decisions on behalf of people, the important question is not only how quickly the system works, but who is responsible when it makes a mistake and whether data subjects have the right to challenge the decision. This makes calls for regulation an everyday concern, not merely an issue discussed on stages by AI executives.
When AI Enters Our Lives but the Rules Cannot Keep Up
Imagine seeing a deepfake video and being unable to tell what is real, or applying for a job only to be rejected by an automated system without any explanation. These problems are not remote or abstract, because they affect the credibility, income, and opportunities of ordinary people.
When companies use AI to make decisions on behalf of people, the important question is not only how quickly the system works, but who is responsible when it makes a mistake and whether data subjects have the right to challenge the decision. This makes calls for regulation an everyday concern, not merely an issue discussed on stages by AI executives.
Where AI of Each Era Stands in the Industry Battleground
In the era of laboratories and specialized models, companies typically focused on research and enterprise customers. Their positions on regulation therefore tended to support safety standards in order to build trust and reduce business risks.
As AI moved toward creative platforms, companies had to manage users, developers, and vast amounts of data. Executives consequently called for clear rules, while also taking care not to make regulations so strict that smaller competitors could no longer keep up.
A company’s position in the industry directly affects its stance. Larger companies are better able to absorb the costs of regulation, while newer companies often worry that the same rules could become barriers to growth.
Where AI of Each Era Stands in the Industry Battleground
In the era of laboratories and specialized models, companies typically focused on research and enterprise customers. Their positions on regulation therefore tended to support safety standards in order to build trust and reduce business risks.
As AI moved toward creative platforms, companies had to manage users, developers, and vast amounts of data. Executives consequently called for clear rules, while also taking care not to make regulations so strict that smaller competitors could no longer keep up.
A company’s position in the industry directly affects its stance. Larger companies are better able to absorb the costs of regulation, while newer companies often worry that the same rules could become barriers to growth.
From Warnings About Risk to More Detailed Proposals
Calls from AI executives gradually shifted from broad warnings to proposals that more clearly define the responsibilities of each party, including model testing, data disclosure, and accountability when harm occurs.
| Factor | Early stage | Later stage |
|---|---|---|
| Nature of the calls | Warning about AI risks in general | Proposing detailed regulatory approaches |
| Model testing | Discussing safety as a principle | Calling for testing before deployment |
| Data disclosure | No clear responsibilities specified | Proposing disclosure of necessary information |
| Role of the state | Calling for greater understanding | Proposing that the state establish rules and provide oversight |
The key turning point is the recognition that safety must be supported by verifiable evidence, not merely warnings from the companies themselves.
From Warnings About Risk to More Detailed Proposals
Calls from AI executives gradually shifted from broad warnings to proposals that more clearly define the responsibilities of each party, including model testing, data disclosure, and accountability when harm occurs.
| Factor | Early stage | Later stage |
|---|---|---|
| Nature of the calls | Warning about AI risks in general | Proposing detailed regulatory approaches |
| Model testing | Discussing safety as a principle | Calling for testing before deployment |
| Data disclosure | No clear responsibilities specified | Proposing disclosure of necessary information |
| Role of the state | Calling for greater understanding | Proposing that the state establish rules and provide oversight |
The key turning point is the recognition that safety must be supported by verifiable evidence, not merely warnings from the companies themselves.
What Does “AI Governance” Mean in Practice?
AI governance means establishing rules that enable oversight before a system is made available for widespread use. Safety must be tested and outcomes monitored after deployment, rather than waiting for problems to occur before addressing them.
Disclosing training data helps with copyright reviews and makes it possible to assess the reliability of outputs. At the same time, external agencies should inspect black-box systems to determine whether they contain risks that companies may not have identified.
Ultimately, responsibilities must be clearly defined. If AI causes harm within an organization, it must be clear who is responsible for prevention, oversight, and the consequences that follow.
What Does “AI Governance” Mean in Practice?
AI governance means establishing rules that enable oversight before a system is made available for widespread use. Safety must be tested and outcomes monitored after deployment, rather than waiting for problems to occur before addressing them.
Disclosing training data helps with copyright reviews and makes it possible to assess the reliability of outputs. At the same time, external agencies should inspect black-box systems to determine whether they contain risks that companies may not have identified.
Ultimately, responsibilities must be clearly defined. If AI causes harm within an organization, it must be clear who is responsible for prevention, oversight, and the consequences that follow.
Who Says What, and Where Are the Interests?
| Factor | AI company executives | Independent researchers | Government |
|---|---|---|---|
| Proposals | Common rules and safety standards | Transparent audits and data disclosure | Enacting laws and assigning responsibility |
| Strengths | Data and resources from real-world development | Ability to ask questions independently | Ability to enforce rules across all companies |
| Limitations | May design rules that are difficult for competitors to follow | Limited enforcement power | Slow processes and difficulty keeping up with technology |
| Business incentives | Reducing risks and building trust | Maintaining research standards | Protecting the public and preserving regulatory authority |
The important issue is therefore not only who is calling for AI to be controlled, but also who benefits from those rules. Good governance should create space for all three parties to scrutinize one another.
Who Says What, and Where Are the Interests?
| Factor | AI company executives | Independent researchers | Government |
|---|---|---|---|
| Proposals | Common rules and safety standards | Transparent audits and data disclosure | Enacting laws and assigning responsibility |
| Strengths | Data and resources from real-world development | Ability to ask questions independently | Ability to enforce rules across all companies |
| Limitations | May design rules that are difficult for competitors to follow | Limited enforcement power | Slow processes and difficulty keeping up with technology |
| Business incentives | Reducing risks and building trust | Maintaining research standards | Protecting the public and preserving regulatory authority |
The important issue is therefore not only who is calling for AI to be controlled, but also who benefits from those rules. Good governance should create space for all three parties to scrutinize one another.
Calls That Reduce Risk Versus Calls That Could Become Tools of Exclusion
AI executives have data from real-world development and use, enabling them to identify risks that laws should address more precisely. However, they must disclose conflicts of interest and allow affected people to participate as well.
If rules are complex or expensive to comply with, large companies will gain an advantage and smaller competitors may find it harder to enter the market. More importantly, discussion of future risks should not obscure problems that already exist, such as using data without consent or the impact on workers.
Pros
- +Helps identify risks from real-world use
- +Aligns rules with the technology
Cons
- −Risks allowing large companies to design rules that favor themselves
- −May divert attention from problems that have already occurred
Calls That Reduce Risk Versus Calls That Could Become Tools of Exclusion
AI executives have data from real-world development and use, enabling them to identify risks that laws should address more precisely. However, they must disclose conflicts of interest and allow affected people to participate as well.
If rules are complex or expensive to comply with, large companies will gain an advantage and smaller competitors may find it harder to enter the market. More importantly, discussion of future risks should not obscure problems that already exist, such as using data without consent or the impact on workers.
Pros
- +Helps identify risks from real-world use
- +Aligns rules with the technology
Cons
- −Risks allowing large companies to design rules that favor themselves
- −May divert attention from problems that have already occurred
The Price Society Pays When Rules Come Too Late
When rules arrive too late, the cost does not end with corporate budgets. It may fall on workers who are replaced, victims of false information, and people whose rights are violated without clear channels for redress.
Those with less access to technology are placed at an even greater disadvantage, while government agencies must spend more money and deploy more personnel to inspect systems, handle complaints, and address problems that have already occurred. In the end, society pays, even though it was not the one that decided to adopt the technology.
The Price Society Pays When Rules Come Too Late
When rules arrive too late, the cost does not end with corporate budgets. It may fall on workers who are replaced, victims of false information, and people whose rights are violated without clear channels for redress.
Those with less access to technology are placed at an even greater disadvantage, while government agencies must spend more money and deploy more personnel to inspect systems, handle complaints, and address problems that have already occurred. In the end, society pays, even though it was not the one that decided to adopt the technology.
Lessons from History That Can Still Guide AI’s Future
Executives’ calls have shifted from warnings about risks to proposals for legal frameworks, standards, and governance mechanisms. Their positions become credible only when they accept responsibility, disclose risk information, and support oversight that also applies to their own companies.
Policymakers should therefore not allow industry players to write the rules alone. They must listen to researchers, workers, affected groups, and independent agencies, while separating business interests from decision-making.
The key question may not be “Should AI be regulated?” but rather “Who should be regulated, and what should they be responsible for?” Developers, users, and organizations that use systems to make decisions on behalf of people should all have clearly defined responsibilities.
Lessons from History That Can Still Guide AI’s Future
Executives’ calls have shifted from warnings about risks to proposals for legal frameworks, standards, and governance mechanisms. Their positions become credible only when they accept responsibility, disclose risk information, and support oversight that also applies to their own companies.
Policymakers should therefore not allow industry players to write the rules alone. They must listen to researchers, workers, affected groups, and independent agencies, while separating business interests from decision-making.
The key question may not be “Should AI be regulated?” but rather “Who should be regulated, and what should they be responsible for?” Developers, users, and organizations that use systems to make decisions on behalf of people should all have clearly defined responsibilities.
From Warnings on Stage to Policy-Level Issues
At first, AI executives typically discussed risks on public stages while calling for rules to govern the technology. Later, the issue moved from warnings based on general principles to debates about standards, transparency, and corporate accountability.
The broader picture therefore involves not only how dangerous AI is, but also how developers should be audited and what remedies should be available to people affected by AI. Calls for governance become meaningful only when they are turned into rules that can actually be enforced.
From Warnings on Stage to Policy-Level Issues
At first, AI executives typically discussed risks on public stages while calling for rules to govern the technology. Later, the issue moved from warnings based on general principles to debates about standards, transparency, and corporate accountability.
The broader picture therefore involves not only how dangerous AI is, but also how developers should be audited and what remedies should be available to people affected by AI. Calls for governance become meaningful only when they are turned into rules that can actually be enforced.
When AI Enters Our Lives but the Rules Cannot Keep Up
One morning, we may see a fake video that misleads people, only to discover later that its images and audio were generated by AI. At work, an automated system may screen applications or make decisions without providing an explanation that ordinary people can understand.
The problem is not remote at all, because the results may directly affect our jobs, income, or reputations. When AI executives call for governance rules, this is therefore not merely a matter for technology companies. It raises the question of who is responsible when systems fail and how affected people can request an explanation.
When AI Enters Our Lives but the Rules Cannot Keep Up
One morning, we may see a fake video that misleads people, only to discover later that its images and audio were generated by AI. At work, an automated system may screen applications or make decisions without providing an explanation that ordinary people can understand.
The problem is not remote at all, because the results may directly affect our jobs, income, or reputations. When AI executives call for governance rules, this is therefore not merely a matter for technology companies. It raises the question of who is responsible when systems fail and how affected people can request an explanation.
Where AI of Each Era Stands in the Industry Battleground
In the early days, AI was still confined to laboratories or companies developing specialized models. Executives therefore tended to view regulation as a matter of safety and professional standards, rather than rules that directly affected the market.
As AI became a creative platform, model-owning companies gained revenue, reputation, and influence over large numbers of users. Their positions on regulation consequently became more complex: rules could build trust, but they could also raise costs and give competitors an opportunity to catch up.
Business position therefore clearly shapes the tone of executives. Those concerned about risks call for oversight, while those focused on rapid market expansion emphasize rules that do not obstruct innovation.
Where AI of Each Era Stands in the Industry Battleground
In the early days, AI was still confined to laboratories or companies developing specialized models. Executives therefore tended to view regulation as a matter of safety and professional standards, rather than rules that directly affected the market.
As AI became a creative platform, model-owning companies gained revenue, reputation, and influence over large numbers of users. Their positions on regulation consequently became more complex: rules could build trust, but they could also raise costs and give competitors an opportunity to catch up.
Business position therefore clearly shapes the tone of executives. Those concerned about risks call for oversight, while those focused on rapid market expansion emphasize rules that do not obstruct innovation.
From Warnings About Risk to More Detailed Proposals
At first, AI executives often spoke broadly about risks, such as threats to society, work, and safety. Their statements therefore sounded serious, but did not clearly explain how these risks should be managed.
Later, the calls became more detailed, ranging from testing models before deployment to disclosing data, assigning responsibility, and asking the state to establish a regulatory framework.
| Factor | Early stage | Later stage |
|---|---|---|
| Nature of the warnings | Abstract warnings of danger | Identifying risks that must be examined |
| Model testing | No detailed requirements | Testing before deployment |
| Accountability | Discussing safety in general | Assigning responsibility when problems occur |
| Role of the state | Calling for awareness of the problem | Proposing a clear governance framework |
From Warnings About Risk to More Detailed Proposals
At first, AI executives often spoke broadly about risks, such as threats to society, work, and safety. Their statements therefore sounded serious, but did not clearly explain how these risks should be managed.
Later, the calls became more detailed, ranging from testing models before deployment to disclosing data, assigning responsibility, and asking the state to establish a regulatory framework.
| Factor | Early stage | Later stage |
|---|---|---|
| Nature of the warnings | Abstract warnings of danger | Identifying risks that must be examined |
| Model testing | No detailed requirements | Testing before deployment |
| Accountability | Discussing safety in general | Assigning responsibility when problems occur |
| Role of the state | Calling for awareness of the problem | Proposing a clear governance framework |
What Does “AI Governance” Mean in Practice?
Before releasing AI for widespread use, organizations must test whether the system produces dangerous answers or discriminates, because errors can affect users on a broad scale.
Organizations should disclose as much training data as possible to address copyright issues and help assess how reliable the outputs are.
Critical systems should also be reviewed by external agencies, especially when users cannot explain how the AI reached its decisions.
If AI harms customers, organizations must clearly identify who is responsible, from the development team and system approvers to the people using it in practice.
What Does “AI Governance” Mean in Practice?
Before releasing AI for widespread use, organizations must test whether the system produces dangerous answers or discriminates, because errors can affect users on a broad scale.
Organizations should disclose as much training data as possible to address copyright issues and help assess how reliable the outputs are.
Critical systems should also be reviewed by external agencies, especially when users cannot explain how the AI reached its decisions.
If AI harms customers, organizations must clearly identify who is responsible, from the development team and system approvers to the people using it in practice.
Who Says What, and Where Are the Interests?
| Factor | AI company executives | Independent researchers / Government |
|---|---|---|
| Proposals | Making rules together with industry | Auditing and setting rules externally |
| Strengths | Understanding the technology and its real costs | Reducing favoritism toward any one company |
| Limitations | May establish rules that favor large companies | Risk falling behind technological change |
| Potential hidden incentives | Creating barriers for new competitors and reducing legal risks | Protecting the public and preserving the credibility of research |
Calls for governance therefore do not necessarily mean that every party is acting solely out of sacrifice for society. We must also consider who benefits from the rules and who writes them.
Who Says What, and Where Are the Interests?
| Factor | AI company executives | Independent researchers / Government |
|---|---|---|
| Proposals | Making rules together with industry | Auditing and setting rules externally |
| Strengths | Understanding the technology and its real costs | Reducing favoritism toward any one company |
| Limitations | May establish rules that favor large companies | Risk falling behind technological change |
| Potential hidden incentives | Creating barriers for new competitors and reducing legal risks | Protecting the public and preserving the credibility of research |
Calls for governance therefore do not necessarily mean that every party is acting solely out of sacrifice for society. We must also consider who benefits from the rules and who writes them.
Calls That Reduce Risk Versus Calls That Could Become Tools of Exclusion
AI executives understand technical risks and business impacts, allowing them to identify more precisely what regulations should address. However, the voices of large companies should not become the sole standard, because this could increase costs to the point that small companies and researchers struggle to compete.
Good rules should be auditable, provide space for multiple parties to participate, and be reviewed as technology changes. Industry should not be allowed to set conditions that give itself an advantage.
Pros
- +Helps identify technical risks directly
- +Aligns rules with real-world use
Cons
- −May increase costs and exclude smaller companies
- −May divert attention from problems that have already occurred
Calls That Reduce Risk Versus Calls That Could Become Tools of Exclusion
AI executives understand technical risks and business impacts, allowing them to identify more precisely what regulations should address. However, the voices of large companies should not become the sole standard, because this could increase costs to the point that small companies and researchers struggle to compete.
Good rules should be auditable, provide space for multiple parties to participate, and be reviewed as technology changes. Industry should not be allowed to set conditions that give itself an advantage.
Pros
- +Helps identify technical risks directly
- +Aligns rules with real-world use
Cons
- −May increase costs and exclude smaller companies
- −May divert attention from problems that have already occurred
The Price Society Pays When Rules Come Too Late
When rules arrive too late, the cost does not end with company budgets. It may fall on workers who are replaced, consumers exposed to false information, and data owners whose information is used without consent.
Those with less access to technology are placed at an even greater disadvantage, while government agencies must spend more money and deploy more personnel to inspect systems, repair damage, and keep up with systems that change faster than the law.
The Price Society Pays When Rules Come Too Late
When rules arrive too late, the cost does not end with company budgets. It may fall on workers who are replaced, consumers exposed to false information, and data owners whose information is used without consent.
Those with less access to technology are placed at an even greater disadvantage, while government agencies must spend more money and deploy more personnel to inspect systems, repair damage, and keep up with systems that change faster than the law.
Lessons from History That Can Still Guide AI’s Future
Calls from AI executives have shifted from requesting “common rules” to acknowledging that companies must also be responsible for the impacts of their own systems. But words alone are not enough. Signs of sincerity include opening data to scrutiny, accepting audits, and supporting standards that apply equally to every company.
Policymakers should clearly separate the roles of rulemakers, auditors, and the companies being regulated. Industry should contribute information, but should not be the sole party responsible for setting the rules.
The key question may therefore not be “Should AI be regulated?” but rather “Who should be regulated, and what should they be responsible for?” Developers, users, and executives who decide to deploy these systems in practice should all have clear responsibilities.
Lessons from History That Can Still Guide AI’s Future
Calls from AI executives have shifted from requesting “common rules” to acknowledging that companies must also be responsible for the impacts of their own systems. But words alone are not enough. Signs of sincerity include opening data to scrutiny, accepting audits, and supporting standards that apply equally to every company.
Policymakers should clearly separate the roles of rulemakers, auditors, and the companies being regulated. Industry should contribute information, but should not be the sole party responsible for setting the rules.
The key question may therefore not be “Should AI be regulated?” but rather “Who should be regulated, and what should they be responsible for?” Developers, users, and executives who decide to deploy these systems in practice should all have clear responsibilities. This article traces the history of calls for AI regulation, from early warnings by executives to current proposals for laws and standards, while examining why voices from industry leaders carry both influence, contradictions, and conflicts of interest.
The key point is that these calls are not only about safety. They also concern business direction, the power to set rules, and social responsibility.
This article traces the history of calls for AI regulation, from early warnings by executives to current proposals for laws and standards, while examining why voices from industry leaders carry both influence, contradictions, and conflicts of interest.
The key point is that these calls are not only about safety. They also concern business direction, the power to set rules, and social responsibility.
From Warnings on Stage to Policy-Level Issues
Calls for AI regulation began with warnings from executives on public stages before evolving into proposals concerning laws, standards, and corporate accountability. Technology is therefore no longer viewed merely as a product, but as something that affects society as a whole.
As the issue entered the policy arena, the influence of executives grew alongside questions about conflicts of interest, because the same people may both call for rules and have a stake in those rules.
From Warnings on Stage to Policy-Level Issues
Calls for AI regulation began with warnings from executives on public stages before evolving into proposals concerning laws, standards, and corporate accountability. Technology is therefore no longer viewed merely as a product, but as something that affects society as a whole.
As the issue entered the policy arena, the influence of executives grew alongside questions about conflicts of interest, because the same people may both call for rules and have a stake in those rules.
When AI Enters Our Lives but the Rules Cannot Keep Up
Imagine seeing a deepfake video and being unable to tell what is real, or applying for a job only to be rejected by an automated system without any explanation. These problems are not remote or abstract, because they affect the credibility, income, and opportunities of ordinary people.
When companies use AI to make decisions on behalf of people, the important question is not only how quickly the system works, but who is responsible when it makes a mistake and whether data subjects have the right to challenge the decision. This makes calls for regulation an everyday concern, not merely an issue discussed on stages by AI executives.
When AI Enters Our Lives but the Rules Cannot Keep Up
Imagine seeing a deepfake video and being unable to tell what is real, or applying for a job only to be rejected by an automated system without any explanation. These problems are not remote or abstract, because they affect the credibility, income, and opportunities of ordinary people.
When companies use AI to make decisions on behalf of people, the important question is not only how quickly the system works, but who is responsible when it makes a mistake and whether data subjects have the right to challenge the decision. This makes calls for regulation an everyday concern, not merely an issue discussed on stages by AI executives.
Where AI of Each Era Stands in the Industry Battleground
In the era of laboratories and specialized models, companies typically focused on research and enterprise customers. Their positions on regulation therefore tended to support safety standards in order to build trust and reduce business risks.
As AI moved toward creative platforms, companies had to manage users, developers, and vast amounts of data. Executives consequently called for clear rules, while also taking care not to make regulations so strict that smaller competitors could no longer keep up.
A company’s position in the industry directly affects its stance. Larger companies are better able to absorb the costs of regulation, while newer companies often worry that the same rules could become barriers to growth.
Where AI of Each Era Stands in the Industry Battleground
In the era of laboratories and specialized models, companies typically focused on research and enterprise customers. Their positions on regulation therefore tended to support safety standards in order to build trust and reduce business risks.
As AI moved toward creative platforms, companies had to manage users, developers, and vast amounts of data. Executives consequently called for clear rules, while also taking care not to make regulations so strict that smaller competitors could no longer keep up.
A company’s position in the industry directly affects its stance. Larger companies are better able to absorb the costs of regulation, while newer companies often worry that the same rules could become barriers to growth.
From Warnings About Risk to More Detailed Proposals
Calls from AI executives gradually shifted from broad warnings to proposals that more clearly define the responsibilities of each party, including model testing, data disclosure, and accountability when harm occurs.
| Factor | Early stage | Later stage |
|---|---|---|
| Nature of the calls | Warning about AI risks in general | Proposing detailed regulatory approaches |
| Model testing | Discussing safety as a principle | Calling for testing before deployment |
| Data disclosure | No clear responsibilities specified | Proposing disclosure of necessary information |
| Role of the state | Calling for greater understanding | Proposing that the state establish rules and provide oversight |
The key turning point is the recognition that safety must be supported by verifiable evidence, not merely warnings from the companies themselves.
From Warnings About Risk to More Detailed Proposals
Calls from AI executives gradually shifted from broad warnings to proposals that more clearly define the responsibilities of each party, including model testing, data disclosure, and accountability when harm occurs.
| Factor | Early stage | Later stage |
|---|---|---|
| Nature of the calls | Warning about AI risks in general | Proposing detailed regulatory approaches |
| Model testing | Discussing safety as a principle | Calling for testing before deployment |
| Data disclosure | No clear responsibilities specified | Proposing disclosure of necessary information |
| Role of the state | Calling for greater understanding | Proposing that the state establish rules and provide oversight |
The key turning point is the recognition that safety must be supported by verifiable evidence, not merely warnings from the companies themselves.
What Does “AI Governance” Mean in Practice?
AI governance means establishing rules that enable oversight before a system is made available for widespread use. Safety must be tested and outcomes monitored after deployment, rather than waiting for problems to occur before addressing them.
Disclosing training data helps with copyright reviews and makes it possible to assess the reliability of outputs. At the same time, external agencies should inspect black-box systems to determine whether they contain risks that companies may not have identified.
Ultimately, responsibilities must be clearly defined. If AI causes harm within an organization, it must be clear who is responsible for prevention, oversight, and the consequences that follow.
What Does “AI Governance” Mean in Practice?
AI governance means establishing rules that enable oversight before a system is made available for widespread use. Safety must be tested and outcomes monitored after deployment, rather than waiting for problems to occur before addressing them.
Disclosing training data helps with copyright reviews and makes it possible to assess the reliability of outputs. At the same time, external agencies should inspect black-box systems to determine whether they contain risks that companies may not have identified.
Ultimately, responsibilities must be clearly defined. If AI causes harm within an organization, it must be clear who is responsible for prevention, oversight, and the consequences that follow.
Who Says What, and Where Are the Interests?
| Factor | AI company executives | Independent researchers | Government |
|---|---|---|---|
| Proposals | Common rules and safety standards | Transparent audits and data disclosure | Enacting laws and assigning responsibility |
| Strengths | Data and resources from real-world development | Ability to ask questions independently | Ability to enforce rules across all companies |
| Limitations | May design rules that are difficult for competitors to follow | Limited enforcement power | Slow processes and difficulty keeping up with technology |
| Business incentives | Reducing risks and building trust | Maintaining research standards | Protecting the public and preserving regulatory authority |
The important issue is therefore not only who is calling for AI to be controlled, but also who benefits from those rules. Good governance should create space for all three parties to scrutinize one another.
Who Says What, and Where Are the Interests?
| Factor | AI company executives | Independent researchers | Government |
|---|---|---|---|
| Proposals | Common rules and safety standards | Transparent audits and data disclosure | Enacting laws and assigning responsibility |
| Strengths | Data and resources from real-world development | Ability to ask questions independently | Ability to enforce rules across all companies |
| Limitations | May design rules that are difficult for competitors to follow | Limited enforcement power | Slow processes and difficulty keeping up with technology |
| Business incentives | Reducing risks and building trust | Maintaining research standards | Protecting the public and preserving regulatory authority |
The important issue is therefore not only who is calling for AI to be controlled, but also who benefits from those rules. Good governance should create space for all three parties to scrutinize one another.
Calls That Reduce Risk Versus Calls That Could Become Tools of Exclusion
AI executives have data from real-world development and use, enabling them to identify risks that laws should address more precisely. However, they must disclose conflicts of interest and allow affected people to participate as well.
If rules are complex or expensive to comply with, large companies will gain an advantage and smaller competitors may find it harder to enter the market. More importantly, discussion of future risks should not obscure problems that already exist, such as using data without consent or the impact on workers.
Pros
- +Helps identify risks from real-world use
- +Aligns rules with the technology
Cons
- −Risks allowing large companies to design rules that favor themselves
- −May divert attention from problems that have already occurred
Calls That Reduce Risk Versus Calls That Could Become Tools of Exclusion
AI executives have data from real-world development and use, enabling them to identify risks that laws should address more precisely. However, they must disclose conflicts of interest and allow affected people to participate as well.
If rules are complex or expensive to comply with, large companies will gain an advantage and smaller competitors may find it harder to enter the market. More importantly, discussion of future risks should not obscure problems that already exist, such as using data without consent or the impact on workers.
Pros
- +Helps identify risks from real-world use
- +Aligns rules with the technology
Cons
- −Risks allowing large companies to design rules that favor themselves
- −May divert attention from problems that have already occurred
The Price Society Pays When Rules Come Too Late
When rules arrive too late, the cost does not end with corporate budgets. It may fall on workers who are replaced, victims of false information, and people whose rights are violated without clear channels for redress.
Those with less access to technology are placed at an even greater disadvantage, while government agencies must spend more money and deploy more personnel to inspect systems, handle complaints, and address problems that have already occurred. In the end, society pays, even though it was not the one that decided to adopt the technology.
The Price Society Pays When Rules Come Too Late
When rules arrive too late, the cost does not end with corporate budgets. It may fall on workers who are replaced, victims of false information, and people whose rights are violated without clear channels for redress.
Those with less access to technology are placed at an even greater disadvantage, while government agencies must spend more money and deploy more personnel to inspect systems, handle complaints, and address problems that have already occurred. In the end, society pays, even though it was not the one that decided to adopt the technology.
Lessons from History That Can Still Guide AI’s Future
Executives’ calls have shifted from warnings about risks to proposals for legal frameworks, standards, and governance mechanisms. Their positions become credible only when they accept responsibility, disclose risk information, and support oversight that also applies to their own companies.
Policymakers should therefore not allow industry players to write the rules alone. They must listen to researchers, workers, affected groups, and independent agencies, while separating business interests from decision-making.
The key question may not be “Should AI be regulated?” but rather “Who should be regulated, and what should they be responsible for?” Developers, users, and organizations that use systems to make decisions on behalf of people should all have clearly defined responsibilities.
Lessons from History That Can Still Guide AI’s Future
Executives’ calls have shifted from warnings about risks to proposals for legal frameworks, standards, and governance mechanisms. Their positions become credible only when they accept responsibility, disclose risk information, and support oversight that also applies to their own companies.
Policymakers should therefore not allow industry players to write the rules alone. They must listen to researchers, workers, affected groups, and independent agencies, while separating business interests from decision-making.
The key question may not be “Should AI be regulated?” but rather “Who should be regulated, and what should they be responsible for?” Developers, users, and organizations that use systems to make decisions on behalf of people should all have clearly defined responsibilities.
From Warnings on Stage to Policy-Level Issues
At first, AI executives typically discussed risks on public stages while calling for rules to govern the technology. Later, the issue moved from warnings based on general principles to debates about standards, transparency, and corporate accountability.
The broader picture therefore involves not only how dangerous AI is, but also how developers should be audited and what remedies should be available to people affected by AI. Calls for governance become meaningful only when they are turned into rules that can actually be enforced.
From Warnings on Stage to Policy-Level Issues
At first, AI executives typically discussed risks on public stages while calling for rules to govern the technology. Later, the issue moved from warnings based on general principles to debates about standards, transparency, and corporate accountability.
The broader picture therefore involves not only how dangerous AI is, but also how developers should be audited and what remedies should be available to people affected by AI. Calls for governance become meaningful only when they are turned into rules that can actually be enforced.
When AI Enters Our Lives but the Rules Cannot Keep Up
One morning, we may see a fake video that misleads people, only to discover later that its images and audio were generated by AI. At work, an automated system may screen applications or make decisions without providing an explanation that ordinary people can understand.
The problem is not remote at all, because the results may directly affect our jobs, income, or reputations. When AI executives call for governance rules, this is therefore not merely a matter for technology companies. It raises the question of who is responsible when systems fail and how affected people can request an explanation.
When AI Enters Our Lives but the Rules Cannot Keep Up
One morning, we may see a fake video that misleads people, only to discover later that its images and audio were generated by AI. At work, an automated system may screen applications or make decisions without providing an explanation that ordinary people can understand.
The problem is not remote at all, because the results may directly affect our jobs, income, or reputations. When AI executives call for governance rules, this is therefore not merely a matter for technology companies. It raises the question of who is responsible when systems fail and how affected people can request an explanation.
Where AI of Each Era Stands in the Industry Battleground
In the early days, AI was still confined to laboratories or companies developing specialized models. Executives therefore tended to view regulation as a matter of safety and professional standards, rather than rules that directly affected the market.
As AI became a creative platform, model-owning companies gained revenue, reputation, and influence over large numbers of users. Their positions on regulation consequently became more complex: rules could build trust, but they could also raise costs and give competitors an opportunity to catch up.
Business position therefore clearly shapes the tone of executives. Those concerned about risks call for oversight, while those focused on rapid market expansion emphasize rules that do not obstruct innovation.
Where AI of Each Era Stands in the Industry Battleground
In the early days, AI was still confined to laboratories or companies developing specialized models. Executives therefore tended to view regulation as a matter of safety and professional standards, rather than rules that directly affected the market.
As AI became a creative platform, model-owning companies gained revenue, reputation, and influence over large numbers of users. Their positions on regulation consequently became more complex: rules could build trust, but they could also raise costs and give competitors an opportunity to catch up.
Business position therefore clearly shapes the tone of executives. Those concerned about risks call for oversight, while those focused on rapid market expansion emphasize rules that do not obstruct innovation.
From Warnings About Risk to More Detailed Proposals
At first, AI executives often spoke broadly about risks, such as threats to society, work, and safety. Their statements therefore sounded serious, but did not clearly explain how these risks should be managed.
Later, the calls became more detailed, ranging from testing models before deployment to disclosing data, assigning responsibility, and asking the state to establish a regulatory framework.
| Factor | Early stage | Later stage |
|---|---|---|
| Nature of the warnings | Abstract warnings of danger | Identifying risks that must be examined |
| Model testing | No detailed requirements | Testing before deployment |
| Accountability | Discussing safety in general | Assigning responsibility when problems occur |
| Role of the state | Calling for awareness of the problem | Proposing a clear governance framework |
From Warnings About Risk to More Detailed Proposals
At first, AI executives often spoke broadly about risks, such as threats to society, work, and safety. Their statements therefore sounded serious, but did not clearly explain how these risks should be managed.
Later, the calls became more detailed, ranging from testing models before deployment to disclosing data, assigning responsibility, and asking the state to establish a regulatory framework.
| Factor | Early stage | Later stage |
|---|---|---|
| Nature of the warnings | Abstract warnings of danger | Identifying risks that must be examined |
| Model testing | No detailed requirements | Testing before deployment |
| Accountability | Discussing safety in general | Assigning responsibility when problems occur |
| Role of the state | Calling for awareness of the problem | Proposing a clear governance framework |
What Does “AI Governance” Mean in Practice?
Before releasing AI for widespread use, organizations must test whether the system produces dangerous answers or discriminates, because errors can affect users on a broad scale.
Organizations should disclose as much training data as possible to address copyright issues and help assess how reliable the outputs are.
Critical systems should also be reviewed by external agencies, especially when users cannot explain how the AI reached its decisions.
If AI harms customers, organizations must clearly identify who is responsible, from the development team and system approvers to the people using it in practice.
What Does “AI Governance” Mean in Practice?
Before releasing AI for widespread use, organizations must test whether the system produces dangerous answers or discriminates, because errors can affect users on a broad scale.
Organizations should disclose as much training data as possible to address copyright issues and help assess how reliable the outputs are.
Critical systems should also be reviewed by external agencies, especially when users cannot explain how the AI reached its decisions.
If AI harms customers, organizations must clearly identify who is responsible, from the development team and system approvers to the people using it in practice.
Who Says What, and Where Are the Interests?
| Factor | AI company executives | Independent researchers / Government |
|---|---|---|
| Proposals | Making rules together with industry | Auditing and setting rules externally |
| Strengths | Understanding the technology and its real costs | Reducing favoritism toward any one company |
| Limitations | May establish rules that favor large companies | Risk falling behind technological change |
| Potential hidden incentives | Creating barriers for new competitors and reducing legal risks | Protecting the public and preserving the credibility of research |
Calls for governance therefore do not necessarily mean that every party is acting solely out of sacrifice for society. We must also consider who benefits from the rules and who writes them.
Who Says What, and Where Are the Interests?
| Factor | AI company executives | Independent researchers / Government |
|---|---|---|
| Proposals | Making rules together with industry | Auditing and setting rules externally |
| Strengths | Understanding the technology and its real costs | Reducing favoritism toward any one company |
| Limitations | May establish rules that favor large companies | Risk falling behind technological change |
| Potential hidden incentives | Creating barriers for new competitors and reducing legal risks | Protecting the public and preserving the credibility of research |
Calls for governance therefore do not necessarily mean that every party is acting solely out of sacrifice for society. We must also consider who benefits from the rules and who writes them.
Calls That Reduce Risk Versus Calls That Could Become Tools of Exclusion
AI executives understand technical risks and business impacts, allowing them to identify more precisely what regulations should address. However, the voices of large companies should not become the sole standard, because this could increase costs to the point that small companies and researchers struggle to compete.
Good rules should be auditable, provide space for multiple parties to participate, and be reviewed as technology changes. Industry should not be allowed to set conditions that give itself an advantage.
Pros
- +Helps identify technical risks directly
- +Aligns rules with real-world use
Cons
- −May increase costs and exclude smaller companies
- −May divert attention from problems that have already occurred
Calls That Reduce Risk Versus Calls That Could Become Tools of Exclusion
AI executives understand technical risks and business impacts, allowing them to identify more precisely what regulations should address. However, the voices of large companies should not become the sole standard, because this could increase costs to the point that small companies and researchers struggle to compete.
Good rules should be auditable, provide space for multiple parties to participate, and be reviewed as technology changes. Industry should not be allowed to set conditions that give itself an advantage.
Pros
- +Helps identify technical risks directly
- +Aligns rules with real-world use
Cons
- −May increase costs and exclude smaller companies
- −May divert attention from problems that have already occurred
The Price Society Pays When Rules Come Too Late
When rules arrive too late, the cost does not end with company budgets. It may fall on workers who are replaced, consumers exposed to false information, and data owners whose information is used without consent.
Those with less access to technology are placed at an even greater disadvantage, while government agencies must spend more money and deploy more personnel to inspect systems, repair damage, and keep up with systems that change faster than the law.
The Price Society Pays When Rules Come Too Late
When rules arrive too late, the cost does not end with company budgets. It may fall on workers who are replaced, consumers exposed to false information, and data owners whose information is used without consent.
Those with less access to technology are placed at an even greater disadvantage, while government agencies must spend more money and deploy more personnel to inspect systems, repair damage, and keep up with systems that change faster than the law.
Lessons from History That Can Still Guide AI’s Future
Calls from AI executives have shifted from requesting “common rules” to acknowledging that companies must also be responsible for the impacts of their own systems. But words alone are not enough. Signs of sincerity include opening data to scrutiny, accepting audits, and supporting standards that apply equally to every company.
Policymakers should clearly separate the roles of rulemakers, auditors, and the companies being regulated. Industry should contribute information, but should not be the sole party responsible for setting the rules.
The key question may therefore not be “Should AI be regulated?” but rather “Who should be regulated, and what should they be responsible for?” Developers, users, and executives who decide to deploy these systems in practice should all have clear responsibilities.
Lessons from History That Can Still Guide AI’s Future
Calls from AI executives have shifted from requesting “common rules” to acknowledging that companies must also be responsible for the impacts of their own systems. But words alone are not enough. Signs of sincerity include opening data to scrutiny, accepting audits, and supporting standards that apply equally to every company.
Policymakers should clearly separate the roles of rulemakers, auditors, and the companies being regulated. Industry should contribute information, but should not be the sole party responsible for setting the rules.
The key question may therefore not be “Should AI be regulated?” but rather “Who should be regulated, and what should they be responsible for?” Developers, users, and executives who decide to deploy these systems in practice should all have clear responsibilities.