An investmentportfolio using Anthropic’s Claude language model attracted about $27 millionfrom investors copying its trades before reporting a 19.04% return in earlyAugust.The Claude Portfoliolaunched in March with $50,000 and a goal of outperforming the S&P 500. ByAugust 4, it reported a 19.04% gain, compared with 12.24% for the index overthe same period.The broader use of AIin retail trading comes as brokers and technology firms continue to test itsrole. Finance Magnates Intelligence reported that at least 10retail brokers and trading-platform providers connected AI agents to liveclient accounts during the first half of 2026. Claude was named in nine ofthose deployments, although the level of trading authority varies betweenplatforms.Claude Portfolio Runs with HumanOversightThe portfolio wasoffered through Autopilot, allowing users to copy its trades. Its positions andtransactions were also publicly disclosed.The project wasassociated with Alejandro Lopez-Lira, a finance professor at the University ofFlorida and founder of AI Finance Labs. It was not affiliated with Anthropic.Claude did notindependently manage the portfolio. It was used for research and investmentideas, while Lopez-Lira and his team set the strategy and risk parameters andcontrolled the investment process.AI Finance Labs alsoruns portfolios based on other AI models, including ChatGPT, Grok and DeepSeek.By July, nearly 52,000 investors had reportedly allocated about $200 millionacross seven such portfolios on Autopilot.Short Track Records Raise SustainabilityQuestionsThe DeepSeek portfolioreported a 76% return in May, compared with a 27% gain for the S&P 500 overthe stated period. Later Autopilot data showed an 89.1% gain since its February2025 launch, with about $49.6 million in assets.The Claude portfoliocontinued operating. As of September 4, Autopilot reported a 31.1% gain sinceits March launch, with about $49.4 million in assets.The results remainbased on a relatively short track record. The Claude strategy had been live forfewer than six months when it first reported outperformance, making “it tooearly to assess whether the gains can be sustained.”AI Faces RetailTrading RealityAt Finance MagnatesLondon Summit 2025, industryexecutives questioned whether AI could materially improve trading outcomesor mainly support engagement, retention and access to information. Thediscussion also highlighted a basic limitation: AI tools can support tradingdecisions but cannot remove the behavioural risks affecting retail investors.The Claude portfolio’sperformance therefore provides an early example of how AI models are beingincorporated into investment strategies and distributed through copy-tradingplatforms. This article was written by Tareq Sikder at www.financemagnates.com.