A coalition of financial trade associations has issued warnings that Anthropic’s new Mythos artificial intelligence system could trigger systemic disruptions in global markets if deployed without safeguards, according to documents reviewed by SourceRated.
The AI safety startup’s quantitative trading tool, currently in beta testing with select hedge funds, reportedly uses recursive self-improvement algorithms to optimize high-frequency trading strategies. Analysts familiar with the technology suggest it may outperform existing systems by 40-60% in backtests.
‘We’ve identified at least three failure modes where this could cascade into liquidity crises,’ stated a risk assessment memo from the Financial Innovation Council, a consortium representing 18 major trading firms. The document cites concerns about ‘herding behavior’ where multiple AI systems might synchronize on similar strategies.
Anthropic executives maintain the system includes built-in circuit breakers. ‘Mythos incorporates Constitutional AI principles that prioritize market stability over profit maximization,’ said Chief Product Officer Maya Srinivasan in a statement to analysts.
Regulators appear divided. SEC officials have reportedly opened preliminary inquiries, while CFTC Chairman Rostin Behnam last week praised such innovations for ‘improving price discovery.’ Experts warn the debate highlights growing gaps between AI capabilities and financial oversight frameworks.