Home
Tools
Models
App
Prompts
News
Topics
Submit for inclusion
English
English 日本語 한국어 Português español Deutsch Русский Français 繁體中文 简体中文
option
Home Close
Home
Tools
Category Ranking Generate Cover Generate Image Skill
Models
Large Language Model Multimodal Model
App
Prompts
Prompts Image Prompts
News
News Flash News
Topics
Submit for inclusion
Submit App Submit Tool
Create an account Sign In
Change Language English Change Language
Settings
English EN 日本語 JA 한국어 KO Português PT español ES Deutsch DE Русский RU Français FR 繁體中文 ZH-TW 简体中文 ZH-CN
Home
Flash News
Content
AnthonyRoberts
AnthonyRoberts
October 9, 2026

OpenAI released 722 AI-generated math proofs across 17 fields, sparking debate over cost-efficiency and practical utility. While machine-verified, many proofs lack human interpretability, with single symbol errors causing cascading failures. Mathematicians worry that AI merely recombines existing knowledge, shifting the burden of sorting and interpreting massive outputs onto human researchers without generating the novel insights central to mathematical research.

0
0
Share Click to copy the link
Comments (0)
0/300
Submit
Author JustinScott JustinScott October 9, 2026
OpenAI fired three researchers for allegedly violating security policies, claiming they accessed sensitive data outside procedures. The trio issued an open letter denying misconduct, arguing their collaboration with third-party auditors was authorized to address AI agent risks. They warn that sudden dismissals create a chilling effect, hindering risk monitoring. This dispute highlights the tension between commercial confidentiality and transparent security audits in AI development, raising governance questions over balancing model reasoning capabilities with independent assessments.
0
0
Author DouglasScott DouglasScott October 9, 2026
ByteDance's Seed team identifies phase sensitivity in block-based KV cache compression as the cause of performance fluctuations in large language models processing ultra-long text. This mechanism introduces positional phase coordinates that drastically affect retrieval accuracy, with differences up to 40 percentage points in some open-source models. The findings reveal hidden periodic weaknesses masked by average benchmarks, providing a theoretical basis for optimizing model architecture and enhancing long-context reasoning stability.
0
0
/live/7744
Author HaroldLopez HaroldLopez October 9, 2026
OpenAI launched Ultrafast GPT-6.1Sol across API, Codex, and ChatGPT Work. It runs eight times faster than the standard Sol version but costs six times more at $12 per million input tokens and $60 per million output tokens. Access is granted to Pro500 users, eligible enterprise pay-as-you-go plans, and quota-based education accounts, with enterprise admin activation required. The model is now deployed in all supported regions and supports data residency compliance for the US and EU.
0
0
Author JamesMiller JamesMiller October 9, 2026
Google launched a unified Gemini AI agent system transitioning from Q&A to task execution. With over 1 billion monthly users, it targets enterprise security, featuring goal-oriented planning, deep integration with Workspace and Microsoft 365, and a dedicated digital colleague identity. Early testers include Shopify and PayPal. This move strengthens Google's B-end competitiveness against Microsoft, marking a shift toward advanced autonomous human-machine collaboration in enterprise AI.
0
0
/live/7742
Author BenLewis BenLewis October 8, 2026
Step Terminal unveils STEPX Neo, the first large model-native intelligent agent phone, launching October 13. Featuring Step AOS and AI agent Step Amoo, it executes complex cross-app tasks via single commands, such as generating presentations or ordering deliveries. The device shifts from passive response to active task completion, demonstrating advanced capabilities in photography-assisted scheduling and document handling, marking a significant evolution in smartphone functionality.
0
0
OR