Daily Research Brief 2026-06-11

📊 Token usage: estimated from retrieval and writing scale.

Covers the latest AI research, open source and industry moves, updated daily.

Editor’s Note

The paper thread today: recursive theory-of-mind reasoning, automated prompt injection assessment in agentic environments, a federated learning data-centric review, secure LLM agent threat surfaces, and NEWTON for physically grounded video generation.

1. Latest arXiv Papers

1. Mind the Perspective: Let’s Reason Recursively for Theory of Mind — https://arxiv.org/abs/2606.11724

2. Assessing Automated Prompt Injection Attacks in Agentic Environments — https://arxiv.org/abs/2606.10525

3. Earth-OneVision: Extending Remote Sensing Multimodal LLMs to More Sensor Modalities and Tasks — https://arxiv.org/abs/2606.10819

4. From Data Heterogeneity to Convergence: A Data-Centric Review of Federated Learning — https://arxiv.org/abs/2606.10595

5. Toward Secure LLM Agents: Threat Surfaces, Attacks, Defenses, and Evaluation — https://arxiv.org/abs/2606.10749

6. Why Larger Models Learn More: Effects of Capacity, Interference, and Rare-Task Retention

7. From Context to Skills: Can Language Models Learn from Context Skillfully? — https://arxiv.org/abs/2604.27660

8. NEWTON: Agentic Planning for Physically Grounded Video Generation — https://arxiv.org/abs/2605.18396