Google executive James Manyika has said artificial intelligence companies cannot be solely responsible for policing systems, arguing that regulation and public oversight are necessary alongside industry safeguards.

Manyika, Google’s senior vice-president for Research, Labs, Technology & Society, said responsibility for managing AI’s risks must be shared among developers, governments and society. His comments come as debate intensifies in the United States over whether voluntary commitments by leading technology companies provide sufficient protection.

Speaking in an interview released on Friday, Manyika rejected the idea that concerns about AI should be dismissed as alarmism. Asked about the possibility of catastrophic harm, he said even a 0.1 per cent risk would warrant attention, while stressing that policy should preserve AI’s potential benefits across science, health, education and economic opportunity worldwide, for communities across countries and income levels.

“The dangers have always been real,” Manyika said. “Anytime you’re talking about powerful technologies, we’re all accountable.” He described the appropriate competition in AI as a race to “get it right”, rather than one focused only on accelerating capabilities.

His intervention follows a White House meeting this week at which President Donald Trump and executives from major AI companies signed a voluntary accord covering safety oversight. Google chief executive Sundar Pichai was among the signatories, alongside leaders from Anthropic, Meta, OpenAI, Nvidia and xAI.

The agreement calls for companies to establish robust internal controls, use independent external auditors to assess whether those controls are effective and create board-level committees to review audit findings. The document also says the measures could eventually be incorporated into laws and regulations.

Trump has emphasised self-regulation, saying technology companies understand the need to police themselves, while resisting broad restrictions that he argues could impede US development of AI. He has framed the expansion of the technology as a strategic competition with China and has repeatedly highlighted its expected contribution to investment and economic growth.

Manyika’s position draws a distinction between the responsibilities of companies building frontier systems and the wider governance needed around their deployment. Developers have obligations to test, secure and understand the systems they create, but decisions about acceptable risks, accountability and social consequences extend beyond laboratories.

The argument is consistent with positions Manyika has taken in discussions of AI governance. He has previously said regulation should simultaneously mitigate dangers and enable beneficial applications, rather than treating safety and innovation as mutually exclusive objectives.

That balance has become harder as advanced systems acquire capabilities that can be applied across scientific research, health, education and business, while also creating concerns over cyber security, misinformation, discrimination, privacy and labour disruption. Autonomous systems also raise questions about responsibility when AI tools act with limited human intervention.

Manyika cautioned against allowing public anxiety to overwhelm consideration of AI’s potential value. He said misunderstanding and negative sentiment could obstruct uses capable of benefiting society, placing policymakers and developers under pressure to address risks without closing off applications.

Google has described its AI strategy as both bold and responsible. Its approach includes model evaluations, security testing, governance and AI principles, while the company has supported policy frameworks that impose obligations according to the level of risk presented by applications.

The debate has sharpened as some leading AI executives voice concern about the pace and safety of development. Anthropic chief executive Dario Amodei told the White House gathering that the technology carries “very real risks”, while saying the mechanism for addressing them remains under discussion.

Critics of voluntary oversight argue that promises lack the enforceability, common standards and independent authority associated with statutory regulation. Supporters of a more flexible approach contend that rigid rules can become obsolete quickly in a fast-changing field and may inhibit innovation and safety work.

The White House accord leaves that tension unresolved. It describes its commitments as voluntary and envisages external auditing, while explicitly acknowledging that authorities may later decide to codify some of the safeguards.

Google executive James Manyika has said artificial intelligence companies cannot be solely responsible for policing systems, arguing that regulation and public oversight are necessary alongside industry safeguards.

Manyika, Google’s senior vice-president for Research, Labs, Technology & Society, said responsibility for managing AI’s risks must be shared among developers, governments and society. His comments come as debate intensifies in the United States over whether voluntary commitments by leading technology companies provide sufficient protection.

Speaking in an interview released on Friday, Manyika rejected the idea that concerns about AI should be dismissed as alarmism. Asked about the possibility of catastrophic harm, he said even a 0.1 per cent risk would warrant attention, while stressing that policy should preserve AI’s potential benefits across science, health, education and economic opportunity worldwide, for communities across countries and income levels.

“The dangers have always been real,” Manyika said. “Anytime you’re talking about powerful technologies, we’re all accountable.” He described the appropriate competition in AI as a race to “get it right”, rather than one focused only on accelerating capabilities.

His intervention follows a White House meeting this week at which President Donald Trump and executives from major AI companies signed a voluntary accord covering safety oversight. Google chief executive Sundar Pichai was among the signatories, alongside leaders from Anthropic, Meta, OpenAI, Nvidia and xAI.

The agreement calls for companies to establish robust internal controls, use independent external auditors to assess whether those controls are effective and create board-level committees to review audit findings. The document also says the measures could eventually be incorporated into laws and regulations.

Trump has emphasised self-regulation, saying technology companies understand the need to police themselves, while resisting broad restrictions that he argues could impede US development of AI. He has framed the expansion of the technology as a strategic competition with China and has repeatedly highlighted its expected contribution to investment and economic growth.

Manyika’s position draws a distinction between the responsibilities of companies building frontier systems and the wider governance needed around their deployment. Developers have obligations to test, secure and understand the systems they create, but decisions about acceptable risks, accountability and social consequences extend beyond laboratories.

The argument is consistent with positions Manyika has taken in discussions of AI governance. He has previously said regulation should simultaneously mitigate dangers and enable beneficial applications, rather than treating safety and innovation as mutually exclusive objectives.

That balance has become harder as advanced systems acquire capabilities that can be applied across scientific research, health, education and business, while also creating concerns over cyber security, misinformation, discrimination, privacy and labour disruption. Autonomous systems also raise questions about responsibility when AI tools act with limited human intervention.

Manyika cautioned against allowing public anxiety to overwhelm consideration of AI’s potential value. He said misunderstanding and negative sentiment could obstruct uses capable of benefiting society, placing policymakers and developers under pressure to address risks without closing off applications.

Google has described its AI strategy as both bold and responsible. Its approach includes model evaluations, security testing, governance and AI principles, while the company has supported policy frameworks that impose obligations according to the level of risk presented by applications.

The debate has sharpened as some leading AI executives voice concern about the pace and safety of development. Anthropic chief executive Dario Amodei told the White House gathering that the technology carries “very real risks”, while saying the mechanism for addressing them remains under discussion.

Critics of voluntary oversight argue that promises lack the enforceability, common standards and independent authority associated with statutory regulation. Supporters of a more flexible approach contend that rigid rules can become obsolete quickly in a fast-changing field and may inhibit innovation and safety work.

The White House accord leaves that tension unresolved. It describes its commitments as voluntary and envisages external auditing, while explicitly acknowledging that authorities may later decide to codify some of the safeguards.

Google executive James Manyika has said artificial intelligence companies cannot be solely responsible for policing systems, arguing that regulation and public oversight are necessary alongside industry safeguards.

Manyika, Google’s senior vice-president for Research, Labs, Technology & Society, said responsibility for managing AI’s risks must be shared among developers, governments and society. His comments come as debate intensifies in the United States over whether voluntary commitments by leading technology companies provide sufficient protection.

Speaking in an interview released on Friday, Manyika rejected the idea that concerns about AI should be dismissed as alarmism. Asked about the possibility of catastrophic harm, he said even a 0.1 per cent risk would warrant attention, while stressing that policy should preserve AI’s potential benefits across science, health, education and economic opportunity worldwide, for communities across countries and income levels.

“The dangers have always been real,” Manyika said. “Anytime you’re talking about powerful technologies, we’re all accountable.” He described the appropriate competition in AI as a race to “get it right”, rather than one focused only on accelerating capabilities.

His intervention follows a White House meeting this week at which President Donald Trump and executives from major AI companies signed a voluntary accord covering safety oversight. Google chief executive Sundar Pichai was among the signatories, alongside leaders from Anthropic, Meta, OpenAI, Nvidia and xAI.

The agreement calls for companies to establish robust internal controls, use independent external auditors to assess whether those controls are effective and create board-level committees to review audit findings. The document also says the measures could eventually be incorporated into laws and regulations.

Trump has emphasised self-regulation, saying technology companies understand the need to police themselves, while resisting broad restrictions that he argues could impede US development of AI. He has framed the expansion of the technology as a strategic competition with China and has repeatedly highlighted its expected contribution to investment and economic growth.

Manyika’s position draws a distinction between the responsibilities of companies building frontier systems and the wider governance needed around their deployment. Developers have obligations to test, secure and understand the systems they create, but decisions about acceptable risks, accountability and social consequences extend beyond laboratories.

The argument is consistent with positions Manyika has taken in discussions of AI governance. He has previously said regulation should simultaneously mitigate dangers and enable beneficial applications, rather than treating safety and innovation as mutually exclusive objectives.

That balance has become harder as advanced systems acquire capabilities that can be applied across scientific research, health, education and business, while also creating concerns over cyber security, misinformation, discrimination, privacy and labour disruption. Autonomous systems also raise questions about responsibility when AI tools act with limited human intervention.

Manyika cautioned against allowing public anxiety to overwhelm consideration of AI’s potential value. He said misunderstanding and negative sentiment could obstruct uses capable of benefiting society, placing policymakers and developers under pressure to address risks without closing off applications.

Google has described its AI strategy as both bold and responsible. Its approach includes model evaluations, security testing, governance and AI principles, while the company has supported policy frameworks that impose obligations according to the level of risk presented by applications.

The debate has sharpened as some leading AI executives voice concern about the pace and safety of development. Anthropic chief executive Dario Amodei told the White House gathering that the technology carries “very real risks”, while saying the mechanism for addressing them remains under discussion.

Critics of voluntary oversight argue that promises lack the enforceability, common standards and independent authority associated with statutory regulation. Supporters of a more flexible approach contend that rigid rules can become obsolete quickly in a fast-changing field and may inhibit innovation and safety work.

The White House accord leaves that tension unresolved. It describes its commitments as voluntary and envisages external auditing, while explicitly acknowledging that authorities may later decide to codify some of the safeguards.