Here's what next as Anthropic's most powerful AI model leaked via unsecured data cache - CoinDesk

Tech Share Here's what next as Anthropic's most powerful AI model leaked via unsecured data cache A draft blog post left in an unsecured data cache revealed a new model tier called Capybara that Anthropic says is more capable than anything it has built, with the company flagging "unprecedented" cybersecurity risks. By Shaurya Malwa|Edited by Stephen Alpher Mar 28, 2026, 9:00 a.m. Make preferred on What to know: A leaked draft blog post revealed that Anthropic is testing a new AI model, Claude Mythos, which it internally describes as by far its most powerful system to date. The model, part of a new "Capybara" tier, reportedly delivers dramatically better performance than Claude Opus 4.6 on coding, academic reasoning and cybersecurity tasks, raising concerns about unprecedented cyber risks. The incident underscores both the dual-use danger of advanced AI in areas like DeFi security and the irony that a company touting cutting-edge cybersecurity capabilities exposed details of its own model through a basic content management error. Anthropic is testing the most powerful AI model it has ever built, and the world wasn't supposed to know yet. A data leak reported by Fortune on Thursday revealed that the AI lab behind Claude has trained a new model called "Mythos," which it internally describes as "by far the most powerful AI model we've ever developed." The model was discovered in a draft blog post left in an unsecured, publicly searchable data cache, alongside nearly 3,000 other unpublished assets, according to cybersecurity researchers who reviewed the material. Anthropic confirmed the model's existence after Fortune's inquiry, calling it "a step change" in AI performance and "the most capable we've built to date." The company said it is being trialed by "early access customers" and acknowledged that a "human error" in its content management system caused the leak. The draft blog post introduced a new model tier called "Capybara," described as larger and more capable than Anthropic's existing Opus models, which were previously its most powerful. "Compared to our previous best model, Claude Opus 4.6, Capybara gets dramatically higher scores on tests of software coding, academic reasoning, and cybersecurity, among others," the draft said. It's the cybersecurity dimension that matters most for the crypto industry. The draft blog post said the model "poses unprecedented cybersecurity risks," a framing that has direct implications for blockchain security, smart contract auditing, and the escalating arms race between attackers and defenders in DeFi. This week alone, Ripple announced an AI-driven security overhaul for the XRP Ledger after an AI-assisted red team uncovered more than 10 vulnerabilities in its 13-year-old codebase. Ethereum launched a dedicated post-quantum security hub backed by eight years of research. And the Resolv stablecoin lost its peg after an attacker exploited a minting contract with no oracle checks and single-key access control, the kind of infrastructure failure that more capable AI tools could potentially identify before an attacker does, or exploit faster than defenders can respond. For the AI token market, the leak raises a different question. Bittensor's decentralized network recently released Covenant-72B, a model that competes with Meta's Llama 2 70B, triggering a 90% rally in TAO and driving subnet tokens to a combined market cap of $1.47 billion. A "step change" from a centralized lab like Anthropic resets the benchmark that decentralized AI projects need to match. The competitive distance between what a well-funded corporate lab can build and what a permissionless network can produce just got wider. Anthropic said it is "being deliberate" about the model's release given its capabilities. The draft blog noted the model is expensive to run and not yet ready for general availability. The company removed public access to the data cache after Fortune contacted it. The leak itself is its own cautionary tale. A company building what it describes as an AI model with unprecedented cybersecurity capabilities left the announcement of that model in an unsecured, publicly searchable data store due to human error. The irony needs no elaboration. More For You The Definitive Stablecoin Landscape Series: North America By CoinDesk Research Mar 26, 2026 Commissioned byRipple As stablecoins evolve into core financial infrastructure, North America leads. This report maps the regulation, market shifts, and players driving adoption. Why it matters: Stablecoins are entering their third phase of evolution - the institutionalization era - becoming increasingly embedded into core financial infrastructure. As institutions prioritize transparency and compliance, regulated issuers like USDC, RLUSD, and PYUSD are steadily gaining share with RLUSD surpassing $1B in market cap within its first year. North America, leading in regulatory frameworks and institutional distribution, is at the center of it all. View Full Report Here's how bitcoin, Ethereum and other networks are preparing for the looming quantum threat By Margaux Nijkerk|Edited by Aoyon Ashraf 16 hours ago Across many of the most well-known ecosystems like Bitcoin, Ethereum, and Solana, responses are diverging along familiar lines: what to do on social consensus and technical iteration, and community members are split between caution and acceleration. What to know: The crypto industry is increasingly taking quantum computing seriously, but responses vary widely: Bitcoin is still debating how to act amid concerns over vulnerable coins, while Ethereum and firms like Coinbase are actively building phased, quantum-resistant roadmaps, and Solana is experimenting with optional tools like quantum-safe vaults. Overall, some see... Read Full Story Latest Crypto News Crypto needs a reset before the next bull run 13 hours ago Washington sues Kalshi as states ramp up legal pressure against prediction markets 14 hours ago Kalshi secures license to offer margin trading to institutional investors 15 hours ago Canada moves to ban crypto donations for election campaigns following UK 15 hours ago Why bitcoin's 'compressed' valuation offers reduced downside risk versus stocks 15 hours ago Here's how bitcoin, Ethereum and other networks are preparing for the looming quantum threat 16 hours ago Top Stories Morgan Stanley enters bitcoin ETF race with market-leading low fee Mar 27, 2026 Crypto stocks battered as Nasdaq enters correction in $17 trillion market rout Mar 27, 2026 Bitcoin miners are becoming AI companies and selling their BTC to fund the transition Mar 27, 2026 Anthropic’s massive 'Claude Mythos' leak sends software names — and crypto — sharply lower Mar 27, 2026 Bitcoin drops to two-week low as $300 million in longs are liquidated Mar 27, 2026