One Wrong Reference, One-Year Ban: ArXiv Hallucinated Papers
Thomas G. Dietterich said on 14 May that arXiv authors should face penalties for arxiv hallucinated papers, arguing in an…
-
-
Compute Anxiety, Not Collapse: OpenAI Revenue 2026
OpenAI revenue 2026 is under a real pressure test. In the last 30 days, the dominant story has been a…
-
Claude Code Reasoning Effort Fell, and Quality Followed
Anthropic says the Claude Code reasoning effort default was lowered from high to medium in early March to reduce latency,…
-
Google’s 75% AI-Generated Code Claim Hides the Real Shift
Google says AI-generated code now accounts for 75% of all new code at the company, with that code still “approved…
-
Claude Opus 4.7 Token Usage Jumped 35% Without Price Hikes
Claude Opus 4.7 token usage went up for reasons Anthropic documented in its own launch materials. The list price stayed…
-
Claude Opus 4.6 Support: Docs Say Active, UI Says No
Claude Opus 4.6 support has not been formally dropped. Anthropic’s model deprecations page still lists claude-opus-4-6 as Active and says…
-
Language-Agnostic Representations Show a Shared Semantic Workspace
The standard story is that LLMs work in words. They predict the next token, so surely their internal reasoning is…
-
Speculative Checkpointing Pays Off Only on Repetitive Text
In llama.cpp, speculative checkpointing matters for a simple reason: it points local users toward a cheaper speculative path. You can…
-
The Abstraction Fallacy Makes Conscious AI Harder to Prove
Alexander Lerchner’s paper on conscious AI does something unusual: it does not start by asking whether today’s models seem conscious….
