Anthropic Warns AI Agents Threaten Smart Contract Security

The evidence strongly and consistently supports the truthfulness of the statement. The most compelling evidence comes from two primary sources directly attributed to Anthropic's research team. The first source is explicitly described as the 'direct origin of the warning' and details findings that AI agents can exploit smart contract vulnerabilities. The second primary source is identified as the 'original report' with similar findings. This is corroborated by a high-authority news report which specifically highlights a 'warning' from Anthropic about AI agents executing autonomous attacks on systems that rely on smart contracts. Additional secondary sources, though of lower authority, further amplify and discuss the findings from Anthropic's study, reinforcing the core claim. There is no contradictory evidence presented; the sources that do not support the statement are of low relevance and do not connect the key entities, thus they do not weaken the conclusion. The cumulative weight of direct primary sources and consistent secondary reporting makes it highly probable that Anthropic issued such a warning based on its research.