Anthropic provides Mythos model to EU while blocking AI misuse

Anthropic has implemented several strategic and security measures regarding its artificial intelligence models. The company has provided its advanced Mythos model to the European Union Agency for Cybersecurity (ENISA) to assist in defensive analysis and vulnerability assessment. This distribution excludes the United Kingdom, creating a technical asymmetry between EU regulatory entities and British authorities.

Simultaneously, Anthropic reported that its threat intelligence team has blocked various attempts to misuse its Claude model over the past eight months. These identified activities include attempts to use the AI for propaganda, warfare drones, fraud, and potential biological risks. The company specifically noted influence operations attributed to Russia, China, Iran, and Yemen, including a Russian-linked attempt to generate propaganda disguised as journalism regarding elections in Moldova.