The AI Whistleblower Initiative (AIWI)

If you work at a frontier AI company, you may have noticed something: the systems for raising concerns haven't always kept up with how fast the industry is moving. We think that's worth understanding better. We want to build an evidence base that can help the whole industry develop clear reporting channels, protection guarantees and feedback loops: systems that benefit both employees and the companies they work for. AIWI is an independent nonprofit that works on employee protections in AI. We're conducting the 2026 Frontier AI Workforce Survey to learn what employees across frontier AI companies actually experience when they have concerns. What works? What doesn't? What would help? You may participate in this survey anonymously. We will not ask for your personal details unless you choose to provide them, and we take reasonable steps to limit the collection of metadata that could identify you. We're asking employees from OpenAI, Anthropic, DeepMind, Meta, Microsoft, and xAI to participate. If you'd like to contribute, the basic survey takes just 5 minutes.

We're attempting the largest ever anonymous survey of AI company employees. Want to contribute? The Frontier AI Workforce Survey is an independent research project aiming to understand what frontier AI employees need to feel secure flagging concerns. The survey takes just 5 minutes. We will not ask for your personal details unless you choose to provide them, and we take reasonable steps to limit the collection of metadata that could identify you. We're asking employees from OpenAI, Anthropic, DeepMind, Meta, Microsoft, and xAI to participate. More information can be found on our website.

AIWI is conducting the 2026 Frontier AI Workforce Survey — a research initiative to understand what employees at frontier AI companies need to feel confident raising concerns. Your perspective matters. Please take five minutes to complete the basic survey! You may participate in this survey anonymously. We will not ask for your personal details unless you choose to provide them, and we take reasonable steps to limit the collection of metadata that could identify you. At the link, there will be a QR code for the survey so you can take it on a personal device and network. We're asking employees from OpenAI, Anthropic, DeepMind, Meta, Microsoft, and xAI to participate.

AIWI is a nonprofit that works on employee protections in AI. We've collaborated with regulators across the globe, and our work has been featured in Bloomberg, The Information, and Transformer. Now we're attempting the largest ever survey of AI company employees, the 2026 Frontier AI Workforce Survey: an independent research project to understand what frontier AI employees need to feel secure flagging concerns. Want to contribute? The basic survey takes just 5 minutes. You may participate in this survey anonymously. We will not ask for your personal details unless you choose to provide them, and we take reasonable steps to limit the collection of metadata that could identify you. At the link, there will be a QR code for the survey so you can take it on a personal device and network. We're asking employees from OpenAI, Anthropic, DeepMind, Meta, Microsoft, and xAI to participate.

AIWI is a nonprofit that works on employee protections in AI. We've collaborated with the EU AI Office, and our work has been featured in Bloomberg, The Information, and Transformer. Now we're attempting the largest ever survey of AI company employees, the 2026 Frontier AI Workforce Survey: an independent research project to understand what frontier AI employees need to feel secure flagging concerns. Want to contribute? The basic survey takes just 5 minutes. You may participate in this survey anonymously. We will not ask for your personal details unless you choose to provide them, and we take reasonable steps to limit the collection of metadata that could identify you. At the link, there will be a QR code for the survey so you can take it on a personal device and network. We're asking employees from OpenAI, Anthropic, DeepMind, Meta, Microsoft, and xAI to participate.

Meta recognizes the profound risks of advanced AI systems. But without a published whistleblowing policy, how can employees fully trust reporting channels? Transparency shows genuine commitment to protecting those who speak up. Research shows companies primarily detect misconduct through employees who feel safe reporting it. Published protections help you and your colleagues confidently raise concerns about risks only you can see. Question why Meta should keep these policies secret.