Jeffrey Dastin

Jeffrey Dastin is a technology and business journalist at Reuters, one of the world's largest news agencies, where he has covered the technology industry with a focus on major technology companies, AI, and the intersection of business and policy. He has been based in the San Francisco Bay Area, covering Silicon Valley and the broader technology sector.

Dastin became internationally known for his October 2018 investigation revealing that Amazon had developed and then quietly scrapped an AI-based hiring tool because the system had learned to penalize resumes that contained the word "women's," a direct consequence of training the model on a decade of predominantly male hiring decisions at the company. The investigation documented that Amazon's engineers had discovered the bias and tried to address it, but ultimately concluded the system could not be made neutral and the project was abandoned.

The story became one of the most widely cited cases of AI bias in hiring systems and was instrumental in bringing the problem of biased training data to mainstream public and regulatory attention. It demonstrated in concrete, documentable terms the mechanism that researchers like Cathy O'Neil had described theoretically: AI systems trained on biased human decisions do not discover objective patterns; they learn and scale the biases embedded in their training data.

Dastin has continued to cover AI development, regulation, and business at Reuters, reporting on the major AI companies and the evolving regulatory landscape. His ongoing reporting has examined safety practices and decisions at frontier AI labs, the relationships between major AI companies and government, and the operational realities of deploying AI inside large enterprises. He has been a contributor to investigative pieces on AI safety, on internal disagreements at major AI companies regarding model release decisions, and on the deployment of AI in workplace surveillance and customer-facing systems.

Dastin's reporting style emphasizes documentary evidence: internal documents, on-the-record sources, and verifiable timelines. That methodology is what makes his reporting useful as evidence in a treatise: the cases he documents are not anonymous anecdotes or speculative scenarios but reconstructed industrial practice, sourced and corroborated. His contribution to AI First Principles is the journalistic grounding that connects the framework's claims to the specific industrial behaviors that make those claims necessary.

Jeffrey Dastin provided the most consequential piece of journalism in the AI bias literature. His 2018 Reuters investigation of Amazon's abandoned AI recruiting tool is cited directly in the AI First Principles treatise as the primary evidence case for AI Inherits Messiness: "Amazon's recruiting AI, trained on a decade of predominantly male hiring decisions, learned to penalize resumes that included the word 'women's' — a direct translation of past bias into algorithmic policy."

That single story demonstrated, in a way no academic paper could, that AI bias in training data is not a theoretical concern or a research-lab edge case. It is a documented, deployed, consequential reality that affects real hiring decisions at one of the world's largest employers. Dastin's contribution to the principles is evidentiary: he reported the proof.