92% of AI breaches traced to weak access controls, IBM finds
IBM research reveals most AI-related cyber incidents stem from poor access management, not flaws in the AI itself. What this means for enterprise security.
Artificial intelligence
GSK bets $110M on AI-driven biological data to speed drug discovery
GSK partners with Relation Therapeutics in a $110M deal to generate large-scale biological datasets for AI drug discovery, merging wet-lab experiments with machine learning.
Alibaba’s Qwen3.8-Max: a 2.4T-parameter AI model breaks new ground
Alibaba’s Qwen team has released Qwen3.8-Max, a 2.4-trillion-parameter MoE model with 1M-token context, now broadly available via API and open weights next week.
Supabase’s open benchmark pits AI coders against real tasks
Supabase open-sources Evals, a benchmark that scores Claude Code, Codex and OpenCode on real Supabase projects, then shares the results publicly.
DeepSeek’s V4-Flash-0731 lands with agentic and coding boosts
DeepSeek has released DeepSeek-V4-Flash-0731, a re-trained version of its open MoE model that doubles down on coding and agentic tasks while keeping the same architecture and parameter count. The update is available on Hugging Face and in public beta on the official API, with new support for the Responses format and Codex compatibility.
EU bets €30 billion on AI factories as US tech giants overshadow with $600B
The European Commission plans seven AI gigafactories with €30 billion, while U.S. tech giants plan $600 billion in infrastructure spend this year. What it means for the continent’s ambitions.
Nous Research integrates Hermes Agent with Nostr-based Buzz workspace
Nous Research has added Hermes Agent support for Buzz, Block’s open-source Nostr workspace where humans and AI agents share the same signed channels. Three integration paths are now available, with deployment options ranging from solo developers to enterprises.
OpenAI slashes AI model prices by up to 80% amid China price war
OpenAI’s biggest price cut yet arrives as competition from Chinese providers and Microsoft intensifies.
Can AI spark revolutions? Language models hit a wall
Google DeepMind researcher Tom Zahavy argues language models lack the cognitive tools to drive scientific breakthroughs—world models could be the answer.
OpenAI’s GPT-5.6 Sol outperforms rivals on ARC-AGI-3 under specific settings
OpenAI’s latest GPT-5.6 Sol model achieves a 38.3% score on ARC-AGI-3 using proprietary API enhancements, surpassing Anthropic’s Opus 5—though only under non-standard conditions.