Why AI researchers are fighting for a seat at the table

Employees at DeepMind in London recently asked Google to recognize two labor unions as their official representatives. Google declined the request but agreed to meet via a neutral third-party mediator. The first meeting was arguably a failure; union organizers were frustrated because no senior company leaders attended, leaving only human resources staff to manage the discussion. The workers claim the company is actively trying to discourage them from talking to each other by reconfiguring internal chat spaces. Google maintains that it is engaging in the process neutrally and continues to offer other ways for staff to share their views.

In the tech industry, a union seeks the right to collective bargaining. This is a formal legal process where the employer is required to negotiate terms of employment—such as pay, hours, or, in this case, internal company policy—with a group of employees rather than individuals. For many AI researchers, their work is guided by internal ethics principles. These are lists of promises, like pledging not to build autonomous weapons or tools for mass surveillance. When a company changes these policies, as Alphabet—Google’s parent company—recently did, it changes the fundamental nature of the work the staff is being asked to do. The employees are essentially trying to turn their personal ethical objections into a collective, enforceable contract that the company cannot simply ignore.

The tension reveals a deeper transition in the tech industry. For years, AI labs framed themselves as academic-style research centers focused on solving big problems like complex protein structures. Now, these tools are being integrated into national security and classified government networks. This has created a gap between the goals of executives, who see government contracts as a path to growth and influence, and the employees, who often feel they are being asked to pivot from science to surveillance. The outcome of this dispute will show whether workers in the most lucrative field of our time have the power to influence the purpose of their own creations. It is a fundamental question of agency: in the era of powerful AI, who gets to decide what the technology actually does?