AI Daily · Oct 19, 2026 · Monday
BAD INFLUENCE
Anthropic commits shares to a new public benefit fund that grows with company value
Anthropic is committing 5% of its shares to the newly established Anthropic Public Benefit Fund, subject to stockholder approval. The Fund's stake rises by an additional 2% at the board's discretion for every 50% increase in company value above a set baseline. The Fund is structured as a wholly-owned LLC and will not use its equity for lobbying or political giving. Anthropic says roughly 20% of its shares will be dedicated to the public good once combined with prior co-founder pledges and its employee equity donation match program.
Product launches & updates
meng shao@shao__mengXObsidian CEO open-sources Pele, a zero-dependency Mermaid renderer
Obsidian CEO kepano has open-sourced Pele under the MIT license, a Mermaid diagram renderer that rewrites Mermaid 12.1.0 syntax and outputs SVG. The author reports a 230 kB gzip bundle versus 1.6 MB for mermaid.js and 0.05–0.3 ms render times versus 6–82 ms. Pele has no direct runtime dependencies, compared with 23 for mermaid.js, and will become Obsidian's built-in renderer in its next version.
Papers & research
meng shao@shao__mengX1 more sourceGoogle Research open-sources Harness², a recursive agent harness optimization framework
Google Research has open-sourced Harness², a framework that recursively optimizes agent harnesses for open-world tasks without ground-truth verifiers. It uses Propose, Probe and Compose steps comparing behavior across candidate harnesses, and a persistent library of evidence-backed lessons. The post reports gains of up to +11.5 on WorkBuddy-Bench and +11.8 pp on LAB for GLM-5 with Claude Opus 4.8 as editor, and says the harnesses transfer across models.
Tips & takes
Dongxi NLP@dongxi_nlpXCan AI systems create the next step in their own progress?
Dongxi NLP walks through recursive self-improvement, from a coding agent that rose from 20% to 50% on a 200-task SWE-bench subset by editing its own tools, to five research routes. The author uses the Open-Science workbench to track sources, figures and review, and says the economic models suggest current feedback is not yet enough for self-sustaining acceleration.