Google's AI Safety Team Bypasses Own Hiring AI Systems

Google's AI researchers distrust their own hiring algorithms

Google DeepMind's AGI Safety and Alignment Team has quietly implemented a workaround to its parent company's AI-powered hiring systems, offering job candidates a direct path to human reviewers through a special application form.

The team, which focuses on mitigating risks from advanced artificial intelligence, acknowledged in an internal document that Google's standard application process carries "a non-trivial probability" that qualified candidates' resumes "will be screened out incorrectly or take too long to reach us." The document, marked with instructions not to share widely, was viewed by Bloomberg.

By filling out the alternative form, candidates can ensure "a real human on the team will get to see your application," according to the document.

Why it matters

The revelation exposes a fundamental tension in enterprise AI adoption: even teams building safety measures for artificial intelligence lack confidence in automated systems handling high-stakes decisions like hiring. When AI researchers—people who understand these systems' capabilities and limitations better than most—create workarounds to avoid their own company's AI tools, it raises questions about whether these technologies are ready for widespread deployment in HR departments.

Google's response and broader context

A Google DeepMind spokesperson disputed the characterization that the company's systems filter applicants incorrectly, stating the team created the form to bypass recruiter review and deliver resumes directly to team members. The spokesperson emphasized that "there are no shortcuts to getting hired" and that Google DeepMind aims to recruit the most qualified talent.

The irony is particularly sharp given Google's aggressive marketing of AI hiring tools to corporate clients. The company's Workspace division promotes AI features that can "save HR time by quickly creating drafts for job postings, evaluating resumes, and forecasting hiring needs."

Growing scrutiny of AI hiring systems

Artificial intelligence has rapidly infiltrated recruitment processes, with some HR professionals expressing concern about their own job security. Implementation varies widely—some companies use AI models to rank candidates, while others scan resumes for specific keywords.

These systems face increasing legal and ethical scrutiny. Bloomberg's own investigation found evidence of potential bias in OpenAI's ChatGPT based on applicant names. Workday, a major workplace management software provider, is defending against a lawsuit alleging its AI hiring tools discriminate based on race, age, and disability. Workday has denied the allegations and maintains that humans make final hiring decisions.

The Google DeepMind team's decision to circumvent standard AI screening processes adds another data point to mounting questions about algorithmic hiring's reliability and fairness.

These details were first reported by Bloomberg.