self play
30 articles about self play in AI news
Ctx2Skill: Self-Play Framework Lets LMs Discover Skills Without Labels
Ctx2Skill discovers skills from context via multi-agent self-play without labels. Outputs plug into any LM, targeting manual prompt engineering bottlenecks.
Claude Opus 4.7 Builds AlphaZero-Style Self-Play on Consumer Hardware
Claude Opus 4.7 built AlphaZero self-play from scratch on consumer hardware in three hours, showing autonomous algorithmic code generation.
Beyond Self-Play: The Triadic Architecture for Truly Self-Evolving AI Systems
New research reveals why AI self-play systems plateau and proposes a triadic architecture with three key design principles that enable sustainable self-evolution through measurable information gain across iterations.
AI Teaches Itself to See: Adversarial Self-Play Forges Unbreakable Vision Models
Researchers propose AOT, a revolutionary self-play framework where AI models generate their own adversarial training data through competitive image manipulation. This approach overcomes the limitations of finite datasets to create multimodal models with unprecedented perceptual robustness.
MiniMax Open-Sources M2.7 Model, Details 'Self-Evolution' Training
Chinese AI firm MiniMax has open-sourced its M2.7 model. The key detail from its blog is a 'self-evolution' training process, likened to AlphaGo's self-play, for iterative improvement.
Stanford Public Course Puts Self-Improving AI Agents in Open
Stanford publicly released CS 329S Lecture 1 on self-improving AI agents, taught by Chowdhery and Mirhoseini. The course centers verifier-based sampling and names verification as the key unsolved challenge.
BAAI's AREX: Recursively Self-Improving Research Agents
BAAI releases AREX models that recursively self-improve by alternating research and constraint verification.
LongStraw Reaches 2.1M Tokens on 8 H20 GPUs via Branch Replay
LongStraw reaches 2.1M token positions for RL post-training on 8 H20 GPUs by replaying short response branches, cutting compute 8-16x vs prior art.
Rich Sutton Launches Oak Lab to Build Self-Learning AI Agents
Rich Sutton founded Oak Lab to build self-learning AI agents. He rejects static datasets for real-time reinforcement learning with a trillion-parameter goal at 20W.
Gustaf Westman Pop-Up Opens at Selfridges Through July 25
Gustaf Westman pop-up at Selfridges Corner Shop, open until July 25, showcasing colorful, chunky designs.
OpenAI Codex Record & Replay: One-Shot Workflow Recording Becomes Reusable Skill
OpenAI's Record & Replay lets Codex learn a workflow from one demo and repeat it autonomously. The feature is blocked in the EU, UK, and Switzerland.
Karpathy Joins Anthropic to Lead Recursive Self-Improvement Team
Andrej Karpathy joins Anthropic to lead a new recursive self-improvement team using Claude to accelerate pretraining, per @kimmonismus. The move signals a bet on synthetic data loops over brute-force scaling.
Anthropic Launches Self-Hosted Sandboxes and MCP Tunnels at London Event
Anthropic launched self-hosted sandboxes (public beta) and MCP tunnels (research preview) at Code with Claude London on March 4, 2026, per @bcherny.
Pylon: Self-Host Your Own AI Agent Pipeline That Fixes Sentry Errors via
Pylon is a self-hosted daemon that triggers sandboxed Claude Code agents from webhooks (Sentry, cron, chat) and reports results with human approval — no data leaves your machine.
ID Privacy Launches 'Self-Healing' AI Graph for Automotive Retail
ID Privacy has launched the Self-Healing Agentic Intelligence Graph, an AI platform for automotive retail that automatically updates customer profiles and handles dealer communications. This represents a move towards more autonomous, context-aware AI agents in a high-value retail sector.
Nous Research's Hermes Agent Features Self-Improving Skills, Persistent Memory
A new evaluation of Nous Research's Hermes Agent highlights its self-improving ability to build reusable tools from experience and a smarter persistent memory system that conserves token usage. The agent reportedly improves with continued use, representing a shift towards more adaptive AI systems.
SLSREC: A New Self-Supervised Model for Disentangling Long- and Short-Term User Interests in Recommendations
A new arXiv preprint introduces SLSREC, a self-supervised model that disentangles long-term user preferences from short-term intentions using contrastive learning and adaptive fusion. It outperforms state-of-the-art models on three benchmark datasets, addressing a core challenge in dynamic user modeling.
The Self Driving Portfolio: Agentic Architecture for Institutional Asset Management
Researchers propose an 'agentic strategic asset allocation pipeline' using ~50 specialized AI agents to forecast markets, construct portfolios, and self-improve. The system is governed by a traditional Investment Policy Statement, aiming to automate high-level asset management.
Meta's Hyperagents Enable Self-Referential AI Improvement, Achieving 0.710 Accuracy on Paper Review
Meta researchers introduce Hyperagents, where the self-improvement mechanism itself can be edited. The system autonomously discovered innovations like persistent memory, improving from 0.0 to 0.710 test accuracy on paper review tasks.
DST: Domain-Specialized Tree of Thought Cuts Computational Overhead by 26-75% with Plug-and-Play Predictors
Researchers introduce DST, a plug-and-play predictor that guides Tree of Thought reasoning with lightweight supervised heuristics. The method matches or exceeds standard ToT accuracy while reducing computational costs by 26-75% across mathematical and logical reasoning benchmarks.
Market Report: Key Players and Competitive Dynamics in Computer Vision for Retail
A new market report segments the global computer vision for retail market by component, deployment, retail type, application, and end-user. It highlights competitive dynamics among key players driving adoption in areas like customer analytics and inventory management.
How Large Language Models 'Counter Poisoning': A Self-Purification Battle Involving RAG
New research explores how LLMs can defend against data poisoning attacks through self-purification mechanisms integrated with Retrieval-Augmented Generation (RAG). This addresses critical security vulnerabilities in enterprise AI systems.
Biological Computing Breakthrough: Human Neurons Play DOOM in Petri Dish
Cortical Labs has successfully trained 200,000 human brain cells to play the classic video game DOOM, marking a significant leap toward Synthetic Biological Intelligence. This biological computing approach could solve AI's massive energy consumption problem while enabling new forms of adaptive learning.
Anthropic's 'Cowork Skill' Ushers in New Era of AI Self-Improvement
Anthropic has released a groundbreaking AI 'Cowork Skill' that enables Claude to create and evaluate other AI skills autonomously. This development represents a significant leap toward self-improving AI systems that can benchmark performance and conduct capability interviews.
Solaris: The First Multiplayer World Model That Could Revolutionize Game AI
Researchers have unveiled Solaris, the first multiplayer video world model for Minecraft that generates consistent multi-view observations across multiple players simultaneously. This breakthrough in AI game environments could transform how we build interactive virtual worlds.
PixVerse's 'Playable Reality': AI Blurs Lines Between Video, Games and Virtual Worlds
PixVerse introduces 'Playable Reality,' an AI-generated medium that defies traditional categorization. Blending elements of video, gaming, and virtual environments, this technology creates interactive, dynamic experiences rather than static content.
AI's New Frontier: How Self-Improving Models Are Redefining Machine Learning
Researchers have developed a groundbreaking method enabling AI models to autonomously improve their own training data, potentially accelerating AI development while reducing human intervention. This self-improvement capability represents a significant step toward more autonomous machine learning systems.
The End of the Objective Function? New AI Framework Proposes Self-Regulating Learning Without Goals
Researchers propose a radical departure from traditional AI training, introducing a 'stress-gated' system where AI learns by monitoring its own internal health rather than optimizing external goals. This could enable truly autonomous systems that self-assess and adapt without human supervision.
Beyond Catastrophic Forgetting: AI Research Pioneers Self-Regulating Neural Architectures
Two breakthrough papers introduce Non-Interfering Weight Fields for zero-forgetting learning and objective-free learning systems that self-regulate based on internal dynamics. These approaches could fundamentally change how AI models acquire and retain knowledge.
OpenSage: The Dawn of Self-Programming AI Agents That Build Their Own Teams
OpenSage introduces the first agent development kit enabling LLMs to autonomously create AI agents with self-generated architectures, toolkits, and memory systems, potentially revolutionizing how AI systems are designed and deployed.