Останні новини та аналітика про штучний інтелект | Yellow.com

Ten years ago, a crypto exploit required a rare set of skills. You needed deep Solidity instincts, a working knowledge of bridge architecture and enough patience to trace months of on-chain activity. In 2026, a frontier language model can do the first two in an afternoon and a motivated attacker can

Anthropic released Claude Opus 4.7, a highly capable publicly available artificial intelligence model in the world, by the most closely watched benchmarks in the industry. It claims to write a better code than any competing model. It reasons at near-human expert level across almost every domain tes

Anthropic's new Claude Mythos Preview model is raising alarms across the crypto industry, with Coinbase weighing adoption despite warnings that it could accelerate exchange hacks. Mythos Capabilities Alarm Crypto Anthropic confined Mythos Preview to restricted access through a program called Project

Anthropic has released Claude Opus 4.7, a model that narrows the gap to its restricted Mythos while layering on tighter safety guardrails. Opus 4.7 Launch Details The release slots between Anthropic's standard production models and Mythos Preview, the research system the company declined to ship pub

AI agents now handle roughly 19% of all on-chain activity, a new report says, though full autonomy in decentralized finance remains out of reach. Agent Activity Surges More than 17,000 agents have launched since 2025, according to figures cited in the new DWF Ventures report. Bots already generate o