Coinbase’s CEO, Brian Armstrong, has announced an ambitious goal to increase the percentage of AI-generated code at the cryptocurrency exchange to 50% by October. Currently, AI contributes roughly 40% to the daily code production at Coinbase, reflecting a growing trend across the tech industry to adopt AI coding assistants. Armstrong highlighted the need for responsible use of AI in coding, emphasizing that while AI-generated code holds significant promise, it must still be reviewed and understood by developers.
In a recent tweet, Armstrong stated, “Obviously it needs to be reviewed and understood, and not all areas of the business can use AI-generated code. But we should be using it responsibly as much as we possibly can.” He has been a vocal proponent of integrating AI into the company’s work processes, even admitting to having terminated programmers who were resistant to utilizing AI coding tools. Although he referred to this approach as “heavy-handed,” he maintains that the future of development at Coinbase lies increasingly in AI.
This surge in AI-assisted coding is not unique to Coinbase. A recent report by OpsLevel revealed that 94% of technology companies had staff using AI coding assistants by June 2025, citing productivity and accelerated time-to-market as key advantages. The process of AI-assisted coding involves developers leveraging AI for rapid code generation while still engaging in debugging and oversight of the development process.
However, this trend has been accompanied by the emergence of “vibe coding,” a term coined by computer scientist Andrej Karpathy. Vibe coding means leveraging AI suggestions with minimal oversight, often resulting in developers copying and pasting code without fully understanding it. Karpathy has discussed the phenomenon on social media, referencing how some users interact with AI to generate working code without a deep comprehension of the underlying processes.
Insights from Y Combinator CEO Garry Tan revealed that a significant portion of startups in the Winter 2025 batch has relied heavily on AI, with some reporting that 95% of their code was AI-generated. This growing reliance on AI tools raises concerns about the long-term implications for software developers. Critics warn that constant use of these tools might hinder developers’ understanding of their projects, potentially leading to safety issues as nuances may be overlooked.
Furthermore, questions surrounding the data used to train these AI systems loom large. Art Abal, co-founder of the community-owned data network Vana, expressed unease regarding the origins of AI-generated code. He raised concerns about the value that human-created data brings to the table and whether those who contributed to this data are receiving due recognition or compensation. He stated, “We need a better system of data ownership… Otherwise, humans risk becoming nothing more than ‘data cows’ endlessly milked, never compensated.”
As Coinbase and other technology companies embrace AI, the dialogue around responsible use, oversight, and the ethical implications of AI-generated content continues to evolve, shaping the future landscape of software development and technology deployment.
