Anthropic CEO says values clash with staying relevant

Anthropic CEO Dario Amodei says one of the hardest problems in the emerging AI business is choosing between scale and principles.

In a Bloomberg interview aired earlier this year, Amodei reflected on his time in the industry and argued that many tech companies build around engagement, advertising, and what he described as the spread of low-quality AI content. In that context, he framed the core tension bluntly:

“Either you betray your values, or you become irrelevant.”

The comments came during a discussion about Anthropic’s enterprise-focused products, including Claude Code and Claude Cowork. Amodei suggested that selling tools that are directly useful to businesses is a better fit for the company’s values than pursuing business models optimized for attention at any cost.

That stance is closely tied to Anthropic’s origins. Amodei, a former OpenAI executive, co-founded the company in 2021 in part as a rejection of the values and direction he believed were taking hold at OpenAI at the time. Reports have said he was frustrated by a willingness to bypass safety precautions, including slowing updates to reduce the risk of malicious use.

Recommended reading

Huawei may build its own DRAM empire

Anthropic’s position, though, has drawn criticism of its own. Detractors argue that a company presenting itself as safety-first is still shipping increasingly powerful models, including Mythos, that could create new risks if they fall into the wrong hands.

The backdrop is a broader shift in tech: newer AI-native firms such as OpenAI and Anthropic are widely expected to reshape the industry, with plans to IPO in the coming months potentially challenging the dominance of the “Mag 7” at the top of the US economy.