Google Engineer Rebuts Claims of Lacking Frontier Models
Google DeepMind engineer Logan Kilpatrick has responded to claims that the company is no longer producing cutting-edge AI models. He asserts that Gemini 3.8 Cyber outperforms competitors in several tests.
Google DeepMind engineer Logan Kilpatrick has publicly refuted assertions that the company is no longer at the forefront of AI model development. The claims, made on social media, suggested Google is falling behind, particularly in mathematics and cybersecurity.
Kilpatrick stated that the Gemini 3.8 Cyber AI model has already surpassed competitors like Anthropic's Mythos in numerous cybersecurity tests and practical applications. He noted that internal Google teams within Chrome, Wiz, and Cloud are extensively using the model.
According to Kilpatrick, Google's focus is on enhancing the capabilities of cybersecurity professionals through AI. He indicated that future Gemini 4 series models will continue to advance the development of frontier cybersecurity models.
This exchange occurs amid rapid advancements in AI model development and intense competition among major players such as Google, OpenAI, and Anthropic. Companies are continuously striving to improve the capabilities and security of their models.