Dive into The Register's online archive of incisive tech news reporting, features, and analysis dating back to 1998 ...
GPT-5.3-Codex hits 72.2% exploit success on EVMbench, double its detection rate. AI-enabled exploits cost $1.22 per contract, projected to fall 22% every two months. Crypto absorbs 88% of detected ...
MakinaFi suffered an exploit where it has lost 1,299 ETH. The stolen ETH were swapped to USDC. The swapped USDC coins have been stored in two different wallets. MakinaFi, a DeFi execution engine, has ...
Abstract: Smart contract vulnerabilities continue to cause significant financial losses, despite the implementation of security measures such as manual audits and bug bounty platforms. A critical ...
Non-fungible token lending platform Gondi has vowed to compensate users affected in a Monday exploit during which the attacker stole roughly $230,000 worth of NFTs from the protocol. According to a ...
Gondi said only the Sell & Repay smart contract was affected and that it is safe to continue buying, selling, trading and listing NFTs on the platform. Nonfungible token platform Gondi said it has ...
Yuval Rooz has a blunt message for the smart contract sector: If you claim to be the future plumbing of global finance, you’d better show the cash flow. “People have assigned a lot of value to these ...
Contract analysis once required weeks of manual audits and specialized review, but AI tools can now surface issues in seconds. However, as AI’s capabilities expand ...
EVMbench is the latest collaborative effort between OpenAI and crypto investment firm Paradigm. The tool is designed to measure how well AI agents detect, patch, and exploit vulnerabilities in smart ...
AI agents are now being tested against real smart contract vulnerabilities after $3.4 billion was lost to crypto hacks in 2025 OpenAI introduces EVMbench to test whether AI agents can detect, fix, or ...
OpenAI is stepping deeper into crypto security with the launch of EVMbench, a new testing framework designed to measure how well artificial intelligence can understand and potentially secure smart ...
一些您可能无法访问的结果已被隐去。
显示无法访问的结果