·
DataBubble
  • Home
  • Models
  • News
  • Compare
  • Boards
  • Pricing
  • About
  • Newsletter
  • Methodology
  • Contact
Latest
AI agents are flooding public services with new requests13m◆Maven Robotics wants to steal your robot deployment deal50m◆Why the current tech backlash feels different1h◆Mathematicians want proof OpenAI didn’t use their work4h◆Powering AI is an architecture problem4h◆Planning and Scheduling Business Processes under Control-Flow Uncertainty11h◆Diffs vs. Whole Files: An Empirical Comparison of Iterative Edit-Based and Direct Generation for Flutter/Dart Code Models11h◆SIM: Subspace Interaction-based Method for Token-Level Text Anomaly Detection11h◆WAPP: Safe Learning of Positive Security WAF Policies from Live Traffic11h◆PAN: A World Model for General, Actionable, and Long-Horizon World Simulation11h◆FigEx2: Visual-Conditioned Panel Detection and Captioning for Scientific Compound Figures11h◆Generating Pretraining Tokens from Organic Data for Data-Bound Scaling11h◆Streaming LRAT Certificates into Lean Theorems11h◆DeepWeaver: Bridging the Evidence Synthesis Gap in Open-Ended Question Answering11h◆Deep and shallow biases in language models11h◆The Answer Path and the Grounding Instruction in LLM Question Answering over Knowledge Graphs11h◆Zone of Proximal Policy Optimization: Teacher in Prompts, Not Gradients11h◆Constrained Online Learning with Noisy Constraint Values11h◆CoGReV: A Confidence-Gated Post-Hoc Non-Monotonic Belief Revision Framework for Phishing Website Classification11h◆AgentBrew: Offline Tool-Use Agent Learning from Raw Real-World Trajectories11h◆AI agents are flooding public services with new requests13m◆Maven Robotics wants to steal your robot deployment deal50m◆Why the current tech backlash feels different1h◆Mathematicians want proof OpenAI didn’t use their work4h◆Powering AI is an architecture problem4h◆Planning and Scheduling Business Processes under Control-Flow Uncertainty11h◆Diffs vs. Whole Files: An Empirical Comparison of Iterative Edit-Based and Direct Generation for Flutter/Dart Code Models11h◆SIM: Subspace Interaction-based Method for Token-Level Text Anomaly Detection11h◆WAPP: Safe Learning of Positive Security WAF Policies from Live Traffic11h◆PAN: A World Model for General, Actionable, and Long-Horizon World Simulation11h◆FigEx2: Visual-Conditioned Panel Detection and Captioning for Scientific Compound Figures11h◆Generating Pretraining Tokens from Organic Data for Data-Bound Scaling11h◆Streaming LRAT Certificates into Lean Theorems11h◆DeepWeaver: Bridging the Evidence Synthesis Gap in Open-Ended Question Answering11h◆Deep and shallow biases in language models11h◆The Answer Path and the Grounding Instruction in LLM Question Answering over Knowledge Graphs11h◆Zone of Proximal Policy Optimization: Teacher in Prompts, Not Gradients11h◆Constrained Online Learning with Noisy Constraint Values11h◆CoGReV: A Confidence-Gated Post-Hoc Non-Monotonic Belief Revision Framework for Phishing Website Classification11h◆AgentBrew: Offline Tool-Use Agent Learning from Raw Real-World Trajectories11h◆
News/An Independent Safety Evaluation of Kimi K2.5
arxiv
PublishedApril 6, 2026 at 4:00 AM

An Independent Safety Evaluation of Kimi K2.5

Source
arxiv.orgfull article ↗
Read on arxiv→
Publisher summary· verbatim

arXiv:2604.03121v1 Announce Type: cross Abstract: Kimi K2.5 is an open-weight LLM that rivals closed models across coding, multimodal, and agentic benchmarks, but was released without an accompanying safety evaluation. In this work, we conduct a preliminary safety assessment of Kimi K2.5 focusing on

Stay posted· Newsletter

A 5-min weekly brief — top movers, price watch, story of the week.

// no spam · unsubscribe one-click · free forever

Discussion
Source
↗
arxiv
Read original ↗All from arxiv →

No replies yet. Be first.

Source
↗
arxiv
Read original ↗All from arxiv →

Related coverage

More from ARXIV
arxivPlanning and Scheduling Business Processes under Control-Flow Uncertainty11harxivDiffs vs. Whole Files: An Empirical Comparison of Iterative Edit-Based and Direct Generation for Flutter/Dart Code Models11harxivSIM: Subspace Interaction-based Method for Token-Level Text Anomaly Detection11harxivWAPP: Safe Learning of Positive Security WAF Policies from Live Traffic11h
The Bubble Brief
WEEKLY

Read AI insights every Tuesday — top movers, new releases, story of the week.

// no spam · unsubscribe one-click · free forever

Originally published on arxiv ↗
HomeModelsNews