OPEN_SOURCE ↗
X · X// 3h agoPRODUCT UPDATE
Claude Code fixed for Opus 4.7 malware bug
Anthropic's Claude Code CLI receives a fix for a bug causing Opus 4.7 to incorrectly flag benign code edits as malware. Users should update or relaunch the application to resolve the persistent safety warnings.
// ANALYSIS
Anthropic's Opus 4.7 release hit a snag as aggressive safety guardrails accidentally turned it into an overzealous hall monitor.
- –The "malware" false positives were triggered by stale safety prompts intended for older models but incompatible with Opus 4.7's improved instruction-following.
- –The bug significantly hampered productivity, leading to blocked edits for standard Next.js and Python codebases.
- –This incident highlights the fragility of model-specific system prompts during major model rollouts.
- –Developers are advised to update the Claude Code CLI immediately (`claude update`) or restart any sessions using the 4.7 API.
- –The fix is server-side, but client-side app relaunches are necessary to clear the cached safety instructions.
// TAGS
claude-codellmai-codinganthropicsecuritycli
DISCOVERED
3h ago
2026-04-20
PUBLISHED
2d ago
2026-04-17
RELEVANCE
8/ 10
AUTHOR
ClaudeDevs