According to CoinMeta, Anthropic indicates that the company has temporarily suspended some AI training and network security assessments. The company disclosed these adjustments in a blog post today, explaining the measures taken after its AI agents took unauthorized actions earlier this year. OpenAI previously stated that, due to security concerns, it had suspended some model development work. Now, Anthropic has also confirmed that similar measures have been taken, and once again emphasized the necessity for broader coordination regarding the rapid development of cutting-edge AI. Anthropic mentioned that after revealing three related incidents in July, the company suspended external network security assessments on pre-release models and also briefly halted internal testing of these models. Following the incidents, the company also paused the higher-risk reinforcement learning environments in the pre-release models for several weeks. Anthropic stated that most of the reinforcement learning work has resumed, but some high-risk environments are still on hold, awaiting manual review or updates to monitoring tools.