Legal Advocate Mary Inman: The Next AI Whistleblower Could Come From Anywhere
Key Takeaways
- ▸AI companies use restrictive NDAs, mandatory arbitration clauses, and nondisparagement agreements tied to equity to silence employees, often violating SEC whistleblower protection laws
- ▸Recent high-profile resignations from Anthropic and OpenAI signal growing safety concerns among AI researchers, including the disbanding of OpenAI's mission alignment team
- ▸Nonprofit Psst provides a secure digital safe for anonymous disclosures that matches individual concerns with those of other employees, enabling whistleblowing even in countries without strong protections
Summary
Legal advocate Mary Inman, a founding board member of nonprofit Psst and veteran whistleblower attorney with over 30 years of experience, warns that AI companies are creating a chilling effect on workers who want to speak up about safety concerns. In an interview with Rest of World, Inman—who has represented high-profile whistleblowers including Facebook's Frances Haugen and Theranos's Tyler Shultz—highlighted how restrictive nondisclosure agreements, mandatory arbitration clauses, and nondisparagement agreements are silencing employees at major AI firms.
The comments come amid a wave of recent departures from leading AI companies. Former safety researcher Mrinank Sharma quit Anthropic warning "the world is in peril," while OpenAI researcher Zoë Hitzig resigned over concerns about advertising strategy. OpenAI also reportedly fired a safety executive who raised concerns about erotic content on ChatGPT and disbanded its mission alignment team. Inman notes that the close ties between AI companies and the Trump administration, combined with the industry's immigrant workforce and "AI arms race" rhetoric, have created unprecedented barriers to speaking out.
Psst addresses these challenges through a secure digital safe that allows workers to deposit concerns anonymously and matches them with similar reports from colleagues at the same organization. This approach is particularly valuable for workers in countries without strong whistleblower protections, enabling global participation in AI safety disclosures. Inman emphasized that while impediments are "very high," recent public resignations offer "glimmers of hope" that workers will continue to raise concerns despite the risks.
- The Trump administration's close ties to AI companies and rhetoric around the "AI arms race" with China have created a particularly challenging environment for potential whistleblowers
- The AI industry's heavy reliance on immigrant workers creates additional vulnerability and disincentives for speaking out under current political conditions
