DeepMind, Google’s AI division, has significantly slowed its research publication process to maintain a competitive advantage in the AI industry. Under the leadership of Demis Hassabis, the company has implemented a more rigorous vetting process, making it increasingly challenging for researchers to publish studies. This shift is particularly focused on withholding innovations that could be exploited by competitors or that might negatively reflect on Google’s Gemini AI model. The new policies include a six-month embargo on strategic generative AI papers and require multiple approvals for publication.
While DeepMind continues to produce hundreds of papers annually, the changes have led to frustration among researchers, some of whom have left the organization. The emphasis has shifted from public research contributions to prioritizing product development and commercial interests. Concerns about falling behind in the AI race prompted the merger of DeepMind and Google’s Brain AI unit in 2023, leading to a faster rollout of AI products. However, the increased bureaucracy has stifled the release of potentially sensitive research, causing unrest among staff who value publication in prestigious journals. Overall, the company is navigating the tension between advancing AI technology and safeguarding its competitive edge.
