Claude Code Opus 4.7 reportedly keeps checking on malware, stirring privacy concerns

What users are seeing

It has been reported that users on Hacker News noticed Claude Code Opus 4.7 repeatedly performing malware checks while generating or analyzing code. The thread links to user examples and discussion, and alleges that the model is making persistent safety queries — sometimes more often than expected — during ordinary coding sessions. Developers are surprised and a little unnerved. Is your assistant quietly double-checking everything behind the scenes?

Reactions and risks

Reactions range from amused to alarmed. Some commenters treat it as an overzealous safety layer — "better safe than sorry" — while others worry about telemetry and data leakage when proprietary code is involved. The emotional core here is trust: handing your code to a tool feels different when you suspect it's reaching out for second opinions. Nobody wants an AI that behaves like a nosy QA manager, and yet nobody wants malware-laden output either.

Why this matters now

This episode lands amid ongoing debates about model telemetry, developer privacy, and the trade-offs between safety and confidentiality in code assistants. Major players have been balancing similar tensions — from Copilot to other code LLMs — so it should come as no surprise that users are scrutinizing behavior closely. If the checks are benign safety features, clear disclosure would calm nerves; if not, there are real legal and trust implications.

What's next

For now, details remain largely anecdotal and unconfirmed; it has been reported that there is not yet a clear, authoritative public explanation. Developers are asking for transparency: what gets checked, where it goes, and whether users can opt out. In short, people want the AI to behave like a helpful colleague — not a nosey bystander.

Sources: Hacker News