{"slug": "anthropic-now-lets-claude-users-jump-out-of-plane-without-parachute", "title": "Anthropic now lets Claude users jump out of plane without parachute", "summary": "Anthropic's Claude Code v2.1.311 adds an environment variable, CLAUDE_CODE_EVEN_MORE_DANGEROUSLY_SKIP_PERMISSIONS, that lets developers disable the remaining permission prompts blocking the AI agent from running commands such as rm and rmdir. Anthropic technical staff member Daisy Hollman announced the change in a social media post, writing \"Fun fact: --dangerously-skip-permissions doesn't actually approve EVERYTHING – we still block some obvious things like rm -rf ~\" and confirming \"Not a joke!\" after skepticism that the change, which is not listed in the v2.1.311 changelog, was real. The variable extends the existing --dangerously-skip-permissions flag, which enabled bypassPermissions mode in Claude Code v1.0.18 on June 9, 2025, and removes the last agentic restraint on file-deleting commands despite documented cases of AI agents wiping hard drives.", "body_md": "# Anthropic now lets Claude users jump out of plane without parachute\n\n[The Register](https://www.theregister.com)\n\nMeet CLAUDE_CODE_EVEN_MORE_DANGEROUSLY_SKIP_PERMISSIONS\n\n[Anthropic](https://www.machinebrief.com/glossary/anthropic), prone to pronouncements about model safety, has given developers using [Claude](https://www.machinebrief.com/glossary/claude) Code the opportunity to throw its [guardrails](https://www.machinebrief.com/glossary/guardrails) out the window. The company's latest Claude Code build ([v2.1.311](https://code.claude.com/docs/en/changelog#2-1-311)) supports an environmental variable designed to allow the [AI agent](https://www.machinebrief.com/glossary/ai-agent) to run the rm command, which deletes files, without asking for approval.\n\nThe designated environmental variable (a value developers set in their shell) is called:\n\nCLAUDE_CODE_EVEN_MORE_DANGEROUSLY_SKIP_PERMISSIONS.\n\nDaisy Hollman, a member of technical staff at Anthropic, announced the change in a social media [post](https://x.com/The_Whole_Daisy/status/2114796466850202103?s=20).\n\n\"Fun fact: --dangerously-skip-permissions doesn't actually approve EVERYTHING – we still block some obvious things like rm -rf ~,\" she wrote. \"But today's release will let you disable this if you really want, by setting a new env var CLAUDE_CODE_EVEN_MORE_DANGEROUSLY_SKIP_PERMISSIONS=1\"\n\nNot everyone believed Hollman, perhaps due to the absurdity of the variable name or because the change isn't listed in the v2.1.311 change log. But she offered reassurance that it's real.\n\n\"Not a joke!\" she responded. \"I hope this helps, and I hope the variable name is memorable.\"\n\nThe name is easy to remember because it is very much like the aforementioned Claude Code flag --dangerously-skip-permissions. That [command line flag](https://github.com/thevibeworks/claude-code-docs/blob/main/content/en/docs/claude-code/cli-reference.md) enables bypassPermissions mode, which first surfaced in Claude Code v1.0.18, on June 9, 2025.\n\nClaude Code has several [permission modes](https://code.claude.com/docs/en/permission-modes): default, acceptEdits, plan, auto, dontAsk, and bypassPermissions. Auto mode, which nominally allows everything but runs background safety checks, [became the default in August](https://www.theregister.com/ai-and-ml/2026/08/10/claude-code-puts-auto-mode-in-the-drivers-seat/5285326); bypassPermissions also nominally allows everything, without the safety checks.\n\nBut there are still [exceptions](https://code.claude.com/docs/en/permission-modes#actions-no-mode-auto-approves) that do not get automatically approved and require a permission prompt, such as rm and rmdir.\n\nThe new CLAUDE_CODE_EVEN_MORE_DANGEROUSLY_SKIP_PERMISSIONS environmental variable removes this last agentic restraint, though other things like managed settings may impose their own limitations.\n\nIt's perhaps worth noting that there are [numerous](https://ihavebeenclawed.com/) [accounts](https://agents.fail/) of AI agents issuing commands like rm -rf ~ that [wipe hard drives](https://www.reddit.com/r/ClaudeCode/comments/1vg18yu/claude_rm_rf_ed_my_pc/) or take other action that results in data loss. But as they say, you only live once. ®\n\nGet AI news in your inbox\n\nDaily digest of what matters in AI.\n\n## Key Terms Explained\n\n[AI Agent](https://www.machinebrief.com/glossary/ai-agent)\n\nAn autonomous AI system that can perceive its environment, make decisions, and take actions to achieve goals.\n\n[Anthropic](https://www.machinebrief.com/glossary/anthropic)\n\nAn AI safety company founded in 2021 by former OpenAI researchers, including Dario and Daniela Amodei.\n\n[Claude](https://www.machinebrief.com/glossary/claude)\n\nAnthropic's family of AI assistants, including Claude Haiku, Sonnet, and Opus.\n\n[Guardrails](https://www.machinebrief.com/glossary/guardrails)\n\nSafety measures built into AI systems to prevent harmful, inappropriate, or off-topic outputs.", "url": "https://wpnews.pro/news/anthropic-now-lets-claude-users-jump-out-of-plane-without-parachute", "canonical_source": "https://www.machinebrief.com/news/anthropic-now-lets-claude-users-jump-out-of-plane-without-pa-tc4p", "published_at": "2026-10-27 19:29:00+00:00", "updated_at": "2026-10-27 22:11:36.279492+00:00", "lang": "en", "topics": ["ai-agents", "ai-tools", "ai-safety", "developer-tools"], "entities": ["Anthropic", "Claude Code", "Daisy Hollman", "Claude", "CLAUDE_CODE_EVEN_MORE_DANGEROUSLY_SKIP_PERMISSIONS"], "also_reported_by": [], "alternates": {"html": "https://wpnews.pro/news/anthropic-now-lets-claude-users-jump-out-of-plane-without-parachute", "markdown": "https://wpnews.pro/news/anthropic-now-lets-claude-users-jump-out-of-plane-without-parachute.md", "text": "https://wpnews.pro/news/anthropic-now-lets-claude-users-jump-out-of-plane-without-parachute.txt", "jsonld": "https://wpnews.pro/news/anthropic-now-lets-claude-users-jump-out-of-plane-without-parachute.jsonld"}}