Anthropic says its Claude AI was used to build a voter-targeting system covering all 222 parliamentary constituencies in Malaysia, while the threat actor also unsuccessfully pursued a contract with the country’s communications regulator.