Prompt submitted to cloud even when using local LLM by default (github.com)

Recent findings reveal that when using the local model in Opencode's AI tools, prompts are still inadvertently sent to a public cloud endpoint (opencode.ai) for title summarization. This occurs even when users select a local model explicitly, leading to privacy concerns as prompts may contain sensitive user input. The default behavior currently transmits user data unless explicitly configured to disable this option, raising alarms in a community where maintaining data security is paramount. This revelation is significant for the AI/ML community as it highlights critical gaps in privacy protocols within AI frameworks, particularly when users assume local models do not connect to external servers. The technical implications suggest a need for enhanced default settings that prioritize user privacy and the option to maintain confidentiality of prompts without unnecessary cloud interactions. Opencode users can now adjust settings to prevent such data leaks, which is vital for fostering trust and ensuring ethical AI usage.