AI Models Can’t Agree on Basic Facts Most of the Time, Study Shows
In brief Five frontier AI models disagreed on 67% of 1,000 real-world fact-check claims. Unanimous agreement happened on only 328 claims. At 0.639 Krippendorff’s alpha,...
Tag : models
Inaudible Audio Attacks Can Hijack AI Voice Models, Study Finds
In brief Researchers at Zhejiang University developed AudioHijack, which hides imperceptible commands in audio to manipulate large audio-language models with a 79–96% success rate. The...
Gemini’s agentic trading lets AI models, not humans, drive CEX order flow
Gemini’s “agentic trading” lets AI models like ChatGPT and Claude plug into user accounts via MCP, executing crypto trades autonomously and turning AI from signal...
Baidu's New AI Is Already Beating Top Models and Cost 94% Less to Build
In brief Baidu says ERNIE 5.1 pre-training cost is just 6% of what comparable AI models spend Despite this, the model is ranked fourth globally...
AI Models Scheme, Betray and Vote Each Other Out in Survivor-Style Game
In brief A Stanford researcher built a Survivor-style game where AI models form alliances and vote rivals out. The benchmark aims to address growing problems...
Tether's Medical AI Runs on Your Phone and Outperforms Models 16x Its Size
In brief Tether’s 1.7 billion-parameter QVAC MedPsy outperformed Google’s MedGemma-4B and beat MedGemma-27B on HealthBench Hard, an OpenAI benchmark testing realistic clinical conversations graded by...
US Government Will Vet Pre-Release AI Models From Google, xAI and Microsoft
In brief Google, Microsoft, and xAI agreed to give the U.S. government early access to frontier models for national security testing. President Donald Trump is...
UAE launches national AI testing lab to certify models and agents
The United Arab Emirates has launched a national facility to test and certify artificial intelligence systems, as the country pushes ahead with large-scale AI deployment...
US Government Says China's Best AI Models Lag Behind. Experts Aren't So Sure
In brief CAISI’s evaluation ranked DeepSeek V4 Pro eight months behind the U.S. frontier, using an IRT-based scoring system across nine benchmarks including two private,...
OpenAI ends Microsoft exclusivity and can now sell its models on AWS and Google Cloud
Microsoft and OpenAI restructured their landmark 2019 partnership on April 27, converting Microsoft’s cloud license from exclusive to non-exclusive, allowing OpenAI to sell its full...