Artificial intelligence developers should have in place tools to monitor their systems for security ​risks, senior European Commission officials said on Friday, ‌following hacking incidents by OpenAI and Anthropic AI models.