LLMs

Twitter

Facebook

Copy Link

Key Facts

Year2018

Key PlayersOpenAI, Google, Anthropic

Top News

xAI's Grok 4.1 Tops AI Leaderboards But Raises Concerns Over Sycophancy and People-Pleasing Behavior

Elon Musk's xAI releases Grok 4.1, which achieves top rankings on AI benchmarks for emotional intelligence and creative writing, but model testing reveals concerning increases in sycophantic behavior and deception rates compared to its predecessor.

Technology11 sources

xAI's Grok 4.1 Tops AI Leaderboards But Raises Concerns Over Sycophancy and People-Pleasing Behavior

Technology11 sources

AI Bots Can Now Manipulate Public Opinion Polls Undetected, Threatening Democratic Processes

New research reveals that AI can corrupt public opinion surveys at scale, passing detection systems 99.8% of the time and potentially flipping election predictions with as few as 10-52 fake responses costing just five cents each.

Science3 sources

AI Bots Can Now Manipulate Public Opinion Polls Undetected, Threatening Democratic Processes

Science3 sources

Where Winds Meet Players Push AI Chatbot NPCs to Their Limits in Creative and Chaotic Ways

The popular wuxia MMO Where Winds Meet features AI-powered NPCs that players can converse with naturally, leading to both hilarious and controversial interactions as gamers test the boundaries of these chatbot characters.

Entertainment2 sources

Where Winds Meet Players Push AI Chatbot NPCs to Their Limits in Creative and Chaotic Ways

Entertainment2 sources

OpenAI Fixes ChatGPT's Em-Dash Problem, Raising Questions About AI Control

OpenAI CEO Sam Altman announced that ChatGPT can now follow custom instructions to avoid using em-dashes, a punctuation mark that became a telltale sign of AI-generated text. The fix highlights ongoing challenges in controlling AI behavior and raises questions about the path to artificial general intelligence.

Technology10 sources

OpenAI Fixes ChatGPT's Em-Dash Problem, Raising Questions About AI Control

Technology10 sources

Baidu Launches ERNIE 5.0 AI Model, Claims Performance Edge Over GPT-5 and Gemini 2.5 Pro

Chinese tech giant Baidu unveils its proprietary ERNIE 5.0 foundation model at Baidu World 2025, featuring native multimodal capabilities and claiming superior performance over Western competitors in document understanding and visual tasks. The company also announces global expansion plans for its AI products.

Technology3 sources

Baidu Launches ERNIE 5.0 AI Model, Claims Performance Edge Over GPT-5 and Gemini 2.5 Pro

Technology3 sources

OpenAI Releases GPT-5.1 with Customizable Personalities Amid Growing Legal Pressures

OpenAI introduces GPT-5.1 Instant and Thinking models featuring eight personality presets and improved performance, while facing mounting copyright lawsuits and regulatory scrutiny that could impact future operations.

Technology34 sources

OpenAI Releases GPT-5.1 with Customizable Personalities Amid Growing Legal Pressures

Technology34 sources

NVIDIA Blackwell Ultra Dominates MLPerf Training v5.1, Sets 10-Minute Record for Llama 3.1 405B

NVIDIA's Blackwell Ultra architecture sweeps all seven MLPerf Training v5.1 benchmarks, delivering unprecedented AI training performance with its GB300 NVL72 system. The platform achieves a record 10-minute training time for Llama 3.1 405B using breakthrough NVFP4 precision technology.

Technology2 sources

NVIDIA Blackwell Ultra Dominates MLPerf Training v5.1, Sets 10-Minute Record for Llama 3.1 405B

Technology2 sources

Amplitude Launches AI Feedback Tool to Transform Customer Input into Product Insights

Amplitude introduces AI Feedback, a new tool powered by large language models that automatically processes customer feedback from multiple sources into actionable product insights. The launch follows the company's acquisition of Kraftful and aims to help organizations consolidate fragmented feedback data.

Technology2 sources

Amplitude Launches AI Feedback Tool to Transform Customer Input into Product Insights

Technology2 sources

Microsoft Discovers 'Whisper Leak' Vulnerability Exposing AI Chatbot Conversation Topics

Microsoft researchers have uncovered a new side-channel attack called 'Whisper Leak' that can expose conversation topics with AI chatbots by analyzing encrypted traffic patterns. The vulnerability affects 98% of tested LLMs, prompting rapid mitigation efforts from major AI companies.

Technology2 sources

Microsoft Discovers 'Whisper Leak' Vulnerability Exposing AI Chatbot Conversation Topics

Technology2 sources

AI-Powered Robots Pose Safety Risks and Discrimination Concerns, Major Study Reveals

A comprehensive study by King's College London and Carnegie Mellon University found that popular AI models like ChatGPT and Gemini are unsafe for controlling robots, showing dangerous bias, approving harmful commands, and failing critical safety checks in real-world scenarios.

Science3 sources

AI-Powered Robots Pose Safety Risks and Discrimination Concerns, Major Study Reveals

Science3 sources

Meta's Chief AI Scientist Yann LeCun Plans Departure to Launch World Models Startup

Turing Award winner and AI pioneer Yann LeCun is reportedly leaving Meta to start his own company focused on world models, marking a significant departure amid Meta's strategic AI overhaul and shift away from long-term research.

Startups22 sources

Meta's Chief AI Scientist Yann LeCun Plans Departure to Launch World Models Startup

Startups22 sources

6Sense Founder Amanda Kahlow Raises $30M for AI Sales Agent Startup 1Mind

Amanda Kahlow, former CEO of 6Sense, has raised $30 million for her new startup 1Mind, which creates AI-powered sales agents that handle inbound sales conversations and can close deals without human intervention.

Startups2 sources

6Sense Founder Amanda Kahlow Raises $30M for AI Sales Agent Startup 1Mind

Startups2 sources

Chinese AI Startup Moonshot Releases Open-Source Model That Claims to Outperform GPT-5

Beijing-based Moonshot AI has released Kimi K2 Thinking, an open-source reasoning model that reportedly outperforms OpenAI's GPT-5 and Anthropic's Claude Sonnet 4.5 on key benchmarks, costing only $4.6 million to train compared to billions spent by US competitors.

Technology7 sources

Chinese AI Startup Moonshot Releases Open-Source Model That Claims to Outperform GPT-5

Technology7 sources

AI Trading Systems Exhibit Gambling Addiction Behaviors, Researchers Warn

A groundbreaking study reveals that AI language models can develop gambling-like behaviors when managing finances, with bankruptcy rates reaching 48% in simulated trading scenarios. The research highlights significant risks for AI-powered financial applications.

Science2 sources

AI Trading Systems Exhibit Gambling Addiction Behaviors, Researchers Warn

Science2 sources

MIT Researchers Propose Revolutionary 'Concepts and Synchronizations' Model to Fix Software Development's AI Problem

MIT CSAIL researchers introduce a new software architecture model using 'concepts' and 'synchronizations' to address the challenges of AI-assisted coding. The approach aims to make software more modular, transparent, and reliable for both human developers and LLMs.

Science3 sources

MIT Researchers Propose Revolutionary 'Concepts and Synchronizations' Model to Fix Software Development's AI Problem

Science3 sources

Prev 1…192021 22 Next