AI Safety Bounty Problems Solved Using LLMs and Lean
jd_pressman · x · 2026-08-13
AI safety researcher Wentworth (@BogdanIonutCir2) reports he is now 80% confident that two bounty problems involving natural latents—posted over a year ago with David—have been resolved. Both breakthroughs occurred within the past couple of months and made heavy use of LLMs alongside the Lean theorem prover.
More from Research
- Paper: U.S. Export Controls Unintentionally Accelerated China's Open AI Ecosystems — LuizaJarovsky · 2026-08-13
- LlamaIndex Releases ExtractBench: A Benchmark for Complex Enterprise Document Extraction — llama_index · 2026-08-13
- Fine-tuning Qwen and Gemma with GRPO Beats Claude for News Writing — ivan_bezdomny · 2026-08-13
- Scholar Critiques OpenAI and Anthropic: AI Math Breakthroughs Lack Scientific Rigor — rbhar90 · 2026-08-13
- Dwarkesh Podcast: Every AI Model Has an Inherited Personality — Dwarkesh Patel · 2026-08-13
- New Nature Paper: Decline in Scientific Disruption Was a Data Artefact — EricTopol · 2026-08-13