DeepMind AGI Safety Team Bypasses Google AI Hiring Tools

Google DeepMind's AGI Safety and Alignment Team has created a workaround for its own company's automated hiring software, instructing job candidates to submit a special form to ensure their applications are seen by a human. The move highlights a growing tension in the tech industry, where companies are rapidly adopting AI for recruitment even as concerns mount about the technology's limitations and potential for bias.

According to an internal memo viewed by Bloomberg, the team warned applicants that Google's standard application system carries a "nontrivial probability" that a CV will be screened out incorrectly or take too long to reach the hiring team. The document, which carried the disclaimer "PLEASE DO NOT SHARE THIS DOC WIDELY," advised candidates to fill out an additional form to route their resumes directly to team members. The memo was first reported on August 10, with subsequent coverage appearing on August 11 and 12.

The irony is notable: the AGI Safety and Alignment Team is tasked with mitigating risks from advanced AI, yet it found it necessary to bypass the company's own AI-powered tools for its hiring process. Google markets AI features to corporate clients...