Leak roundup: DeepMind RSI hints, GPT-6 Astra nerf controversy, Kimi K2.8 Code
WorldofAI · youtube · 2026-09-13
This AI news roundup aggregates several rumors and updates (mostly unverified leaks):
- Google DeepMind reportedly exploring recursive self-improvement (RSI): leaked clues point to an internal project called LiveRL, possibly feeding into next-gen Gemini training; Sergey Brin and Demis Hassabis have both signaled AGI ambition.
- GPT-6 Astra quality controversy: reports claim the model was "nerfed" post-launch; OpenAI has responded and fixed the issues.
- Surprise release of Kimi K2.8 Code, with discussion of how it relates to Kimi K3.
- GPT-Live-1 hitting the API.
- Dario Amodei calling for the industry to pace frontier AI development.
Note: the RSI leaks and nerf claims are secondhand rumors of unknown reliability; treat as unconfirmed until official statements.
Related event: Rumors Claim DeepMind Achieved Recursive Self-Improvement(4 posts)→
More from Models
- GPT-6 Astra: 1M-token context, 2.5x price, and a 'Critical' cyber risk rating — Thirumalaivasan_GJ · 2026-09-14
- The jagged frontier of LLM capability is just a map of who built an automated grader — jessi_cata · 2026-09-14
- GPT Image 2.5 is live in ChatGPT, SOTA on all image leaderboards — sherwinwu · 2026-09-14
- Dev Praises Qwen3.8-27B After Week of Testing: One-Shots Vague Prompts, Writes Own Tests — DustNearby2848 · 2026-09-14
- Dev says GLM/DeepSeek one-shots tasks he was paying 'astronomical' prices for — haydendevs · 2026-09-14
- Alexandr Wang: Meta spent a year secretly tuning its models for the Muse personal agent — alexandr_wang · 2026-09-14