Coinbase Raises Concerns About Anthropic's Mythos AI and Bitcoin Confidence

Coinbase expresses concern that Anthropic’s new Claude Mythos AI could exploit vulnerabilities in crypto markets. This assessment emerges in their second-quarter outlook, which suggests that the AI model may autonomously attack weaknesses across various protocols and exchanges. Despite the potential risks posed by AI to the industry, institutional confidence in Bitcoin remains strong, with three-quarters of surveyed investors viewing it as undervalued.

In the context of ongoing struggles in the crypto market, Bitcoin currently trades around $77,000, down nearly 40% from its peak in October, whereas traditional asset classes like the S&P 500 reached new highs in April. David Duong, Coinbase’s head of investment research, notes that many investors are optimistic as speculative trading has diminished, suggesting the market may find stability.

Furthermore, investors have shown renewed interest in Bitcoin exchange-traded funds (ETFs), with approximately $2 billion flowing into these products in April alone, according to data from DefiLlama. Meanwhile, major exchanges are racing to secure access to Mythos, which Anthropic has been cautious about deploying broadly due to its potential risks. The model is capable of identifying and exploiting "zero-day" bugs, which are security flaws unknown to developers, posing a significant threat to sectors like crypto, which heavily depend on online infrastructure and tools for transactions.

Cybersecurity experts highlight the serious implications of AI-enabled attacks, noting that the crypto sector could be particularly vulnerable given its reliance on browsers and open-source applications. In 2025, the industry reportedly lost around $3.4 billion to hacks, underlining the stakes involved as AI capabilities evolve and enter the financial landscape.