Shared Safety Standards Win Majority Support in AGI Poll at 55%

A survey of 100 Americans found broad support for shared safety rules as AI companies compete to build artificial general intelligence. The results show that public acceptance depends on outside oversight, open reporting, and enforceable limits on the race to deploy.

The survey presents a clear challenge for companies pursuing artificial general intelligence, or AGI: technical progress alone will not create public confidence. Respondents remain interested in continued AI development, yet 87.6% chose a slower, more supervised, or halted approach when asked how companies should proceed if one appeared close to reaching AGI.

The strongest support went to a model in which companies continue competing while following shared safety standards. That option received 55.2% of responses, well ahead of every other choice. The result points toward a practical middle ground: private companies keep developing advanced systems, while common rules set a floor for testing, disclosure, monitoring, and release decisions.

The survey asked which competitive structure would inspire the most confidence in AGI development. Respondents selected the following options:

The leading choice combines competition with coordination. That detail matters. Respondents did not select a government-run program as their preferred structure, and they gave little support to one company gaining unchecked control. They favored continued innovation with common safeguards that apply across the industry.

When the cooperative and coordinated options are combined, 70.8% support some form of shared governance. This includes company standards, government coordination, or international management. Purely competitive structures attract 29.2%.

The preference reflects a basic concern about incentives. If companies measure success primarily by who reaches AGI first, safety work can become a cost that slows development. Shared standards can reduce that pressure by giving companies a common set of expectations. A firm that spends more time on testing and control would face less competitive disadvantage if rivals must meet comparable requirements.

A survey of 100 Americans finds cautious support for continued AGI development, but not for an unrestricted race. Respondents most favored companies competing under shared safety standards, while large majorities wanted stronger oversight, prioritized safety alongside or above capability, and assigned release authority outside the developer. The results suggest that transparency, independent evaluation, and enforceable safeguards will shape whether public interest in advanced AI becomes public acceptance. Because the sample was small and not established as probability-based, the findings should be read as directional rather than representative of all Americans.

Respondents expressed nearly equal confidence and skepticism when asked whether leading AI companies would place public safety ahead of the race to develop AGI.

The intensity of concern is important. A small group expressed strong confidence, while a larger group expressed complete distrust. That imbalance creates pressure for companies to demonstrate safety through visible practices and independent review.

The responses also show how confidence changes views about development. Among respondents with higher confidence in AI companies, 53.1% still wanted stronger independent oversight. Only 16% supported moving as quickly as possible. Among respondents with lower confidence, 71.7% wanted development slowed or stopped.

Trust, therefore, functions as a condition for continued development. It does not give companies permission to remove safeguards. Even respondents who feel more confident in the industry generally want external checks before advanced systems reach the public.

The survey found a large preference for safety and control when respondents were asked to choose between the most capable system and the safest, most controllable system.

Taken together, 86.6% placed safety at or above capability. That figure helps explain the rejection of an unrestricted AGI race. People may value better systems, faster research, and useful applications, but they expect control and reliability to keep pace with capability.

For the average user, the technical distinction is straightforward. A more capable system can write, reason, plan, code, or operate tools across a wider range of tasks. A safer and more controllable system gives people greater confidence that it will follow limits, reveal uncertainty, resist misuse, and remain responsive to correction.

Those qualities affect everyday adoption. Employers, schools, hospitals, public agencies, and households all need to know how an advanced system behaves when it encounters an unfamiliar situation. A system that performs impressively in demonstrations may still create serious problems if its failure modes remain hidden or its operators lack effective controls.

The survey’s most important message is not that the public opposes AGI; it is that people want progress separated from unilateral control. Shared standards offer a practical compromise because they preserve competition while limiting the incentive to treat safety as a competitive disadvantage. Companies should view external review and incident disclosure not as obstacles to innovation but as the infrastructure required for trust. A small, non-probability sample cannot settle public opinion, yet the consistency of its direction makes self-certification and speed-first development increasingly difficult to defend.

The question about development speed produced one of the survey’s clearest results. Respondents considered a scenario in which one company appeared close to developing AGI and chose among four possible responses:

The combined result is significant: 87.6% wanted some constraint on the race. The preferred constraint varied, with delay and independent oversight receiving the most support. A small minority supported a complete halt.

The responses show a public willing to tolerate conditional progress. Nearly half, 48.6%, supported continued development either at full speed or under independent oversight. Another 51.4% preferred slowing or stopping. That close split suggests that governance could influence whether people support future development.

A company that publishes testing requirements, accepts external review, and reports serious incidents can give undecided audiences reasons to remain open to progress. A company that treats speed as the central measure of success risks moving those audiences toward delay or prohibition.

Respondents were divided over who should decide when an AGI system is safe enough for release. Government regulators, independent scientific experts, and the company developing the system each received about 29% support. An international regulatory organization received 12.7%.

The distribution still shows a strong preference for outside accountability. About 70.5% assigned the greatest responsibility to a government, independent experts, or an international body. Only 29.5% gave the developer the final say.

That finding challenges self-certification as a governance model. Companies possess important technical knowledge, so they need a role in safety decisions. Their commercial interests also create a conflict when release timing affects revenue, market share, investor expectations, or competitive position.

A workable system could divide responsibilities across several parties:

This layered approach avoids placing every decision with one institution. It also creates multiple points where a rushed or incomplete safety case can receive scrutiny.

The survey separates expectations about technical leadership from confidence in responsible development. OpenAI was viewed as the most likely company to develop AGI first by 31.7% of respondents, yet only 11.4% trusted OpenAI most to develop it responsibly.

That 20.3-point gap is a sharp warning. Many respondents see OpenAI as a serious technical contender while withholding confidence in its governance. Among respondents who predicted that OpenAI would reach AGI first, only 36% also trusted it most to develop the technology responsibly.

Google received a more aligned assessment. Half of respondents viewed Google as the most likely company to develop AGI first, and 54.6% trusted Google most to develop it responsibly. Google was the only company to receive majority support on responsible development.

Those numbers describe public perceptions, not an independent assessment of either company’s actual technical progress or safety performance. Perceptions still matter. They influence regulatory support, employee recruitment, product adoption, partnerships, and the public response to a major release.

The survey also found that 15.1% placed their highest trust in another company. This group showed especially strong skepticism toward the current leaders. Only 4.8% expressed meaningful confidence that leading companies would prioritize safety, and 86.1% wanted development slowed or stopped.

That group may represent a broader demand for institutional alternatives. Some people may trust universities, public research bodies, smaller companies, or future organizations more than the companies currently leading the field. The result adds another argument for independent evaluation and shared standards that apply across the sector.

Attitudes varied by gender, age, and education. Women in the survey expressed substantially lower confidence in companies’ safety priorities than men.

Support for unrestricted speed remained a minority position among men as well. Only 15.2% of male respondents selected that option.

Younger respondents, those ages 18 to 34, were more likely to identify OpenAI as the probable winner and more likely to support rapid development. Even in that group, 53.2% wanted development slowed or stopped.

Education produced a different split in company perceptions. Respondents with bachelor’s or postgraduate degrees were more likely to expect OpenAI to develop AGI first, with 47.9% selecting the company. Respondents with a high school education or less were more likely to select Google for both trust and expected leadership.

These differences can affect how companies communicate about safety. Technical demonstrations may appeal to people who follow AI research closely, while clear explanations of accountability, oversight, and failure reporting may matter more to audiences with less exposure to the companies. Policy makers also need to account for the fact that support for AGI development varies across demographic groups and tends to become more cautious when people are asked about real release decisions.