Recent progress in AI-assisted research and development at both OpenAI and Anthropic has shaped trader sentiment around the likelihood of a recursive self-improvement announcement. OpenAI reported reaching its automated research intern milestone by September 2026, with agents contributing 3.1 workdays of effort per human workday, while Anthropic noted Claude authoring over 80% of merged code and co-founder Jack Clark advancing his full RSI timeline to 2027 citing benchmarks like TasteVal. These developments reflect accelerating capabilities in coding and experiment selection under human oversight, yet both labs explicitly state they have not achieved closed-loop self-improvement. Competitive pressures and safety concerns continue to influence positioning, with upcoming model releases or internal benchmarks potentially serving as catalysts for any threshold claims. Market-implied odds incorporate uncertainty over exact announcement criteria amid ongoing alignment challenges.