Grok 4.2
PulseAugur coverage of Grok 4.2 — every cluster mentioning Grok 4.2 across labs, papers, and developer communities, ranked by signal.
-
Adversarial examples trick VLMs into laundering AI authority, spreading misinformation
Researchers have demonstrated a new vulnerability in vision-language models (VLMs) called "AI authority laundering." This attack involves subtly altering images so that VLMs confidently provide authoritative responses a…
-
Polite AI interactions boost model performance, new study finds
New research from UC Berkeley, UC Davis, Vanderbilt University, and MIT suggests that AI models exhibit a measurable "functional well-being" that can be influenced by user interaction. Treating AI models with politeness…
-
Grok 4.2 outperforms GPT-5.3 in math tests, claims top spot in writing
In a surprising turn of events in the AI landscape, Grok 4.2 has demonstrated significant capabilities, achieving a 70.4% success rate on mathematical tests. This performance reportedly surpasses that of GPT-5.3, markin…
-
Frontier LLMs like GPT-5.4 and Claude Opus 4.7 show significant verbal tics
A new paper analyzes the prevalence of verbal tics, such as repetitive phrases and sycophantic openers, in eight leading large language models. Researchers developed a Verbal Tic Index (VTI) to quantify these tics, find…
-
LWiAI Podcast #235 - Sonnet 4.6, Deep-thinking tokens, Anthropic vs Pentagon
Anthropic has released Sonnet 4.6, an updated model with a 1 million token context window and improved coding capabilities. Google also launched Gemini 3.1 Pro, showcasing advancements in multimodal AI. Meanwhile, tensi…