{"slug": "i-built-a-cryptographic-audit-receipt-for-claude-mythos-and-any-ai-model-here-s", "title": "I built a cryptographic audit receipt for Claude Mythos (and any AI model) — here's how it works", "summary": "A developer named Pulkit has built Aetherproof, an open-source Python library that generates cryptographic audit receipts for AI model outputs in a single function call. The tool creates Ed25519-signed receipts proving when specific findings were generated by Anthropic's Claude Mythos model, and can invisibly embed those receipts into text using Unicode Private Use Area codepoints. In testing, the library passed 187 tests with zero failures and detected all 128-byte flip attempts across 1,000 tamper probes.", "body_md": "Anthropic's Mythos model can autonomously find zero-day\n\nvulnerabilities. Their CVD disclosure process uses manual\n\nSHA-3-512 hash commitments to prove findings existed.\n\nI built something that automates that in one line of Python.\n\nOne function call generates a 128-byte Ed25519-signed receipt\n\nthat proves:\n\n``` python\npython\nimport aetherproof\n\nreceipt = aetherproof.for_anthropic(\n    \"Find vulnerabilities in this binary.\",\n    finding_text,\n    model=\"claude-mythos-preview\"\n)\n\nreceipt.save(\"CVE-2026-001.receipt\")\nprint(receipt.verify())  # True\nTry it in 30 seconds\n\npip install aetherproof\npython -c \"\nimport aetherproof\nr = aetherproof.for_anthropic('question', 'answer')\nprint(r.verify())   # True\nprint(r.pretty())\n\"\nThe unusual part — invisible Unicode watermarking\nReceipts embed invisibly into any text using Unicode\nPrivate Use Area codepoints (U+E000–U+E0FF).\n\nAI output carries its own audit trail. Works in any language —\nArabic, Chinese, Devanagari, Hebrew, Thai, Japanese all tested.\n\nsigned_output = aetherproof.embed(ai_response, receipt.to_bytes())\n# Text looks identical. Receipt is inside.\n\naetherproof.verify_embedded(signed_output)  # True\nNumbers\n187 tests, 0 failures\n128/128 byte flips all detected\n1000/1000 tamper probes pass\nCross-language: Python generates, Rust CLI verifies\n15,446 receipts/sec (Python) · 5,472/sec (Rust)\nWhy AGPL-3.0\nFree for open source. Commercial use needs a license.\nThis is the compliance layer under your AI stack —\nit should be open, auditable, and not vendor-locked.\n\nGitHub\nhttps://github.com/pulkit6732/aetherproof\n\nBuilt by Pulkit. Feedback welcome.\n```\n\n", "url": "https://wpnews.pro/news/i-built-a-cryptographic-audit-receipt-for-claude-mythos-and-any-ai-model-here-s", "canonical_source": "https://dev.to/pulkit_srivastava_8ce4f05/i-built-a-cryptographic-audit-receipt-for-claude-mythos-and-any-ai-model-heres-how-it-works-5hbn", "published_at": "2026-05-29 15:47:13+00:00", "updated_at": "2026-05-29 16:13:42.536513+00:00", "lang": "en", "topics": ["ai-safety", "ai-tools", "ai-products", "ai-infrastructure", "artificial-intelligence"], "entities": ["Anthropic", "Claude Mythos", "Aetherproof", "Ed25519", "SHA-3-512", "CVE-2026-001", "AGPL-3.0", "Rust"], "alternates": {"html": "https://wpnews.pro/news/i-built-a-cryptographic-audit-receipt-for-claude-mythos-and-any-ai-model-here-s", "markdown": "https://wpnews.pro/news/i-built-a-cryptographic-audit-receipt-for-claude-mythos-and-any-ai-model-here-s.md", "text": "https://wpnews.pro/news/i-built-a-cryptographic-audit-receipt-for-claude-mythos-and-any-ai-model-here-s.txt", "jsonld": "https://wpnews.pro/news/i-built-a-cryptographic-audit-receipt-for-claude-mythos-and-any-ai-model-here-s.jsonld"}}