What Anthropic has announced
On July 9, 2026, Anthropic published “Inviting hard questions”, the kickoff of a public campaign inviting people to submit their most critical questions about artificial intelligence. Through the page claude.com/hard-questions, users can view existing questions and add their own – for example, about who decides the rules for AI or whether AI makes the world more dangerous. According to Anthropic, “it’s our job to address them.” The creative agency Mother London conceived the campaign, as reported independently by the trade publication Ad Age.
Anthropic is organized as a Public Benefit Corporation and refers in the announcement to its newly established Anthropic Institute as well as the Long-Term Benefit Trust, which is supposed to oversee compliance with the nonprofit mission – whose board recently welcomed former Fed chief Ben Bernanke. New to the current initiative is the promise to publicly track and document the actions taken in response to the submitted questions, even where Anthropic fails to meet its own goals.
The data basis: Anthropic’s “Public Record”
The campaign is based on Anthropic’s “Public Record”, a survey published on June 12, 2026. According to Anthropic, around 52,000 Americans were surveyed between November and December 2025, supplemented by a separate survey of 81,000 Claude users in 159 countries and 70 languages, as well as several focus groups and evaluations of real usage data. These figures come from Anthropic’s own survey and are independently unverified.
Hopes and concerns of Americans
According to Anthropic, 48 percent of respondents named curing diseases like cancer or Alzheimer’s as one of their three biggest hopes for AI, while 36 percent mentioned support for people with disabilities. Among the concerns, fear of job loss dominated at 64 percent, followed by worries about diminishing independent thinking (56 percent) and misinformation (52 percent). Notably, according to the survey, daily AI users are less likely to worry about their jobs (54 percent) than people who do not use AI (70 percent).
Trust deficit and call for regulation
The survey also shows a significant trust deficit toward the industry: only 15 percent of respondents trust AI companies with development decisions – less than the US federal government (20 percent) and significantly less than independent experts (43 percent). More than 70 percent support government regulation of AI, focusing on data protection (56 percent), child protection (52 percent), and legal liability of companies (49 percent). The measures respondents rated most effective were legal liability for damages (47 percent) and prioritizing safety over growth (44 percent).
Criticism of the initiative
Not all observers view the campaign positively. The analysis platform Resultsense argued even before the launch of the “Hard Questions” campaign that while Anthropic’s survey methodology is solid, the company is making itself the “guardian” of public opinion on AI – precisely at a moment when trust in AI companies is particularly low according to its own figures. The criticism: a lab that finances and evaluates its own survey tends to foreground the topics that fit its preferred solutions. According to Resultsense, the survey largely omits uncomfortable questions about market concentration and the ownership of computing power. This assessment is an independent but likewise externally unverified analysis.
Context
The “Hard Questions” campaign falls into a phase in which Anthropic is increasingly focused on public communication about the societal effects of AI – from Bernanke’s appointment to the Long-Term Benefit Trust to the newly introduced usage overview “Claude Reflect.” Whether the promise to transparently track actions and shortcomings will actually be kept cannot currently be verified. For observers of the AI debate, the initiative remains an example of how a single company is trying to actively shape the public discourse about its own technology.
