AI News on August 11th

AI News 2025/8/11 11:00:10

1. OpenAI's GPT-5 Defeats xAI's Grok 4 in AI Chess Tournament

#OpenAI #GPT5 #xAI

OpenAI's o3 model dominated xAI's Grok 4 in a high-profile AI chess tournament, winning all four matches. The competition required models to play without specialized training, demonstrating raw reasoning capabilities. Chess champion Magnus Carlsen rated both AIs at amateur beginner level (800 ELO).

2. Apple Tests Revolutionary Voice-Only iPhone Control

#Apple #Siri #VoiceAI

Apple is testing a groundbreaking voice control system that eliminates touchscreen dependence. Using AppIntents technology, users can edit/send photos, post social comments, and shop - all through voice commands. The feature could redefine smartphone interaction paradigms when launched.

3. NASA & Google Develop AI Medical Assistant for Astronauts

#NASA #Google #AIMedical

The "CMO-DA" AI assistant achieves 74-88% accuracy diagnosing space-related injuries (back pain, earaches, ankle sprains). Designed for remote space missions, the technology may soon transfer to terrestrial telemedicine applications.

4. OpenAI Reinstates GPT-4o After User Backlash

#OpenAI #GPT4o #UserFeedback

Responding to user dissatisfaction with GPT-5, OpenAI reactivated GPT-4o for Plus/Team subscribers. CEO Sam Altman acknowledged underestimating user attachment to the previous model, highlighting challenges in balancing innovation with user expectations.

5. xAI Makes Grok 4 Free Globally

#xAI #Grok4 #OpenAccess

Elon Musk's xAI removed paywalls for Grok 4, though free users face daily limits. The move aims to expand market share against ChatGPT and Claude, intensifying the AI assistant competition.

6. Meta Acquires AI Voice Startup WaveForms

#Meta #Acquisition #VoiceAI

Meta bolstered its Superintelligence Labs by acquiring WaveForms, founded by ex-OpenAI/Google researchers. The deal accelerates Meta's development of emotional AI and advanced voice models.

7. Claude Opus 4.1 Sets New Coding Benchmark

#Anthropic #Claude #Programming

Anthropic's Claude Opus 4.1 achieves 74.5% accuracy on SWE-bench coding tests while maintaining 98.76% harmless response rate. The update significantly improves code refactoring and error detection capabilities.

8. Google DeepMind Unveils Genie 3 World Model

#Google #Genie3 #WorldModel

DeepMind's Genie 3 generates photorealistic 3D environments in real-time from text prompts, revolutionizing AI training simulations. The model learns physics from video data without traditional engines, enabling dynamic world editing.

9. ElevenLabs Launches Commercial AI Music Generator

#ElevenLabs #AIMusic #Copyright

The voice AI pioneer entered music generation, securing licensing deals with Merlin Network/Kobalt Music. The tool produces commercial-ready tracks from text prompts, navigating copyright challenges through pre-cleared training data.

10. Meta Poaches Apple AI Engineers

#Meta #Apple #TalentWar

Meta recruited at least five Apple AI engineers in recent weeks for its Superintelligence Labs, reflecting intensifying competition for top machine learning talent across Silicon Valley.