News
Claude Mythos and De Fi: Hype or Real Threat?
2+ hour, 45+ min ago (22+ words) Anthropic's Claude Mythos models target vulnerability research; suspension of Claude Fable 5 raises De Fi security fears as AI accelerates smart-contract analysis....
Buterin: Obfuscation Could Enable Private On-Chain Voting
7+ hour, 15+ min ago (17+ words) Vitalik Buterin proposes i O could enable private, collusion-resistant on-chain voting without trusted committees; research needed....
BIP-110 Nears Aug. 7 Signaling Window with 0. 31% Miner Support
1+ day, 21+ hour ago (24+ words) BIP-110 approaches an Aug. 7 signaling window with about 0. 31% miner support for a proposed one-year ban on Ordinals-style data inscriptions....
AI Assistant Resists 6, 000+ Prompt-Injection Emails
2+ day, 23+ hour ago (22+ words) Developer Fernando Irarrazaval's Fiu ignored 6, 000+ prompt-injection emails from 2, 000+ senders; Opus 4. 6 and Open Claw prevented secrets. env leaks....
Story Protocol Rebrands as DATA Foundation, Pivots to AI
4+ day, 2+ hour ago (23+ words) Story Protocol rebrands as DATA Foundation, pivoting from blockchain IP licensing to AI training infrastructure and an on-chain dataset registry called Trace....
Leaked Code Suggests Anthropic May Restore Fable 5
4+ day, 22+ hour ago (25+ words) Strings in Claude Code v2. 1. 190 hint Anthropic may relist Fable 5 and include weekly usage in subscriptions; model remains offline amid export-control talks....
Perfect Fairness in Blockchain Transaction Ordering Is Impossible
5+ day, 2+ hour ago (22+ words) Researchers: perfect fairness in blockchain transaction ordering is impossible-async networks and Condorcet cycles prevent a single receive order and enable MEV....
Bitmine, Sharplink Back Ethlabs to Scale Ethereum for AI Use
5+ day, 14+ hour ago (23+ words) Bitmine and Sharplink funded Ethlabs, a nonprofit by five ex-Ethereum Foundation researchers, to scale Ethereum for AI-driven activity and institutional use....
Qwable: Free Local Model Mimics Fable 5 Reasoning
5+ day, 19+ hour ago (26+ words) Qwable is a free local GGUF model fine-tuned from Qwen3. 6-27 B to mimic Fable 5 reasoning; an ablated variant removes refusal behavior and runs offline....
Z. ai launches GLM-5. 2 on Huawei Ascend chips
1+ week, 3+ day ago (25+ words) Z. ai's open-source GLM-5. 2 (744 B params) trained on Huawei Ascend chips offers 1 M-token context, MIT license, Hugging Face weights and API access....