Maksym Andriushchenko shares ongoing work testing agents on long-horizon benchmarks.
Aug 17, 2026 · 24 posts
3 stories tagged by Digg
Maksym Andriushchenko shares ongoing work testing agents on long-horizon benchmarks.
Aug 17, 2026 · 24 posts
Maksym Andriushchenko posted a thread on the ResearchArena paper evaluating monitors for harmful side tasks in automated AI R&D.
Aug 16, 2026 · 3 posts
A new paper demonstrates large-scale extraction of raw reasoning traces from OpenAI, Anthropic, and Gemini models via API flaws.
Aug 11, 2026 · 102 posts