Descriptions:
Bloomberg Technology interviews Cohere CEO Aidan Gomez days after the US government’s export control action against Anthropic’s Fable 5 and Mythos 5, with Gomez disclosing that Cohere has received a large wave of inbound interest directly attributable to the disruption. Inquiries are coming from enterprise customers seeking to diversify their AI stack, governments outside the US and China concerned about sovereign access to AI infrastructure, and investors reassessing exposure to AI companies ahead of major IPOs.
Gomez frames the ban as a wake-up call for any enterprise that had assumed continuous access to frontier AI models. He argues it accelerates interest in on-premise deployments and models that organizations can fully control — a positioning Cohere has long emphasized with its enterprise and government-focused platform. The interview also covers the technical concept of jailbreaking, with Gomez explaining how safety classifiers designed to prevent misuse can be bypassed through prompt manipulation to unlock the underlying cybersecurity capabilities of a model like Mythos.
On the question of global AI concentration, Gomez acknowledges that only four countries — the US, China, France, and Canada — currently have the capacity to train frontier foundation models, and that foreign-born talent has been essential to US AI progress. He notes that the real fragility may be knowledge concentration in a small number of individuals rather than institutions, and that this expertise has circulated enough across companies that capability diffusion beyond a single lab is already underway.
📺 Source: Bloomberg Technology · Published June 15, 2026
🏷️ Format: Interview
