Google DeepMind’s AGI Safety and Alignment Team faces a peculiar internal bind. Its researchers, tasked with mitigating risks from advanced AI, now steer job candidates around the company’s own automated hiring filters. A memo viewed by Bloomberg reveals the workaround. Applicants receive instructions to submit an extra form. This ensures a human sees their materials.
The document carries a blunt warning. “PLEASE DO NOT SHARE THIS DOC WIDELY.” It states plainly that the applications system carries “a non-trivial probability your CV will be screened out incorrectly or take too long to reach us.” Filling out the form routes resumes directly to team members. Bloomberg first reported the details on August 10.
Google pushes AI tools to clients for sifting job applications at scale. Yet its flagship AI lab distrusts those same systems for its own hires. The safety team works to reduce harms from powerful models. Their memo undercuts confidence in the firm’s HR automation.
A DeepMind spokesperson addressed the matter. The team created the form “to go past the recruiter review, and get their resumes direct to the people on the team.” The company still aims to “recruit and hire the most qualified talent at Google DeepMind.” No shortcuts exist in the process, the spokesperson added. Bloomberg quoted these remarks directly.
Futurism covered the story on August 12. The outlet highlighted the irony for an AI research hub. DeepMind researchers invented a manual bypass for software meant to handle volume. Qualified candidates risk vanishing into the digital void otherwise.
Other outlets echoed the Bloomberg findings. Taipei Times noted Alphabet’s pitch of AI for corporate recruiting. LiveMint detailed the “non-trivial probability” language in the memo. These reports surfaced within the past two days. They draw from the same internal document.
The practice exposes limits in current AI hiring tools. Even sophisticated labs encounter false negatives. Resumes from strong candidates get discarded. Delays compound the issue. Human review becomes essential for critical roles.
Google DeepMind’s careers site lists openings. It emphasizes rigorous, fair processes. Yet the memo suggests automation falls short in practice. Applicants now navigate dual paths. Standard submissions plus the special form.
Such workarounds spread in tech hiring. AI slop floods inboxes. Screening algorithms struggle with volume and nuance. DeepMind’s approach offers one model. Others may follow similar paths.
The episode underscores broader tensions. Firms sell AI efficiency. Internally, they hedge against its flaws. Bloomberg’s reporting captures the memo’s core admission. Futurism frames the contradiction for readers.