·
DataBubble
  • Home
  • Models
  • News
  • Compare
  • Boards
  • Pricing
  • About
  • Newsletter
  • Methodology
  • Contact
Latest
Bridging the Gap: Converting Read Text to Conversational Dialogue14m◆Truth for Believable AI: Expressed Doubt, Provenance, and Belief Revision as an Engineerable Stance14m◆RICE-Alpha: Reliability-Informed Correction with Event Graphs for LLM-Agent Stock Forecasting14m◆The World Won't Wait: Evaluating and Teaching Agents Under a Live Clock14m◆Evaluating Model Capabilities After Benchmark Saturation14m◆Value-Informed Schedule Trimming for Accelerated Sequential Monte Carlo Guidance in Discrete Diffusion14m◆AlphaDiana: A Measurement System for Agents in Verifiable Reasoning Tasks14m◆When Correctness Feedback Erases Evidence: Phase-Separated Completeness Auditing in Verus14m◆Counterfactual Neighborhood Risk Fields: Evidence-Grounded Selective Safety for Vision-Language Models14m◆IneqAgent: Agentic Inequality Proving in Lean with Symbolic Computation14m◆L2G-SR: Learning to Guide LLM-based Symbolic Regression14m◆When Reality Outpaces Cognition: A New Scaling Principle for Reliable Intelligence14m◆ARIA: Runtime Inspection and Amendment for Multi-Agent Systems14m◆Open-world AI: a Tale of Three Requirements14m◆EvolvingBrain: An Adaptive Skill System for Multi-modal Agents Self-Evolution in Open Worlds14m◆When Reasoning Traces Conflict with Evidence: How Answer Configuration Shapes Evidence Following14m◆A Self-Improving Agent that Grows its Own DSL: Oracle-Free Operator Induction for Exact Question Answering14m◆Temporal Specification Guidance for Text-to-Video Generation14m◆Confidence-Calibrated Inference Expansion for Evaluator-Guided Test-Time Reasoning14m◆Evolving Together, Failing Apart: The Skill-Policy Co-evolution Trap in Self-Improving Agents14m◆Bridging the Gap: Converting Read Text to Conversational Dialogue14m◆Truth for Believable AI: Expressed Doubt, Provenance, and Belief Revision as an Engineerable Stance14m◆RICE-Alpha: Reliability-Informed Correction with Event Graphs for LLM-Agent Stock Forecasting14m◆The World Won't Wait: Evaluating and Teaching Agents Under a Live Clock14m◆Evaluating Model Capabilities After Benchmark Saturation14m◆Value-Informed Schedule Trimming for Accelerated Sequential Monte Carlo Guidance in Discrete Diffusion14m◆AlphaDiana: A Measurement System for Agents in Verifiable Reasoning Tasks14m◆When Correctness Feedback Erases Evidence: Phase-Separated Completeness Auditing in Verus14m◆Counterfactual Neighborhood Risk Fields: Evidence-Grounded Selective Safety for Vision-Language Models14m◆IneqAgent: Agentic Inequality Proving in Lean with Symbolic Computation14m◆L2G-SR: Learning to Guide LLM-based Symbolic Regression14m◆When Reality Outpaces Cognition: A New Scaling Principle for Reliable Intelligence14m◆ARIA: Runtime Inspection and Amendment for Multi-Agent Systems14m◆Open-world AI: a Tale of Three Requirements14m◆EvolvingBrain: An Adaptive Skill System for Multi-modal Agents Self-Evolution in Open Worlds14m◆When Reasoning Traces Conflict with Evidence: How Answer Configuration Shapes Evidence Following14m◆A Self-Improving Agent that Grows its Own DSL: Oracle-Free Operator Induction for Exact Question Answering14m◆Temporal Specification Guidance for Text-to-Video Generation14m◆Confidence-Calibrated Inference Expansion for Evaluator-Guided Test-Time Reasoning14m◆Evolving Together, Failing Apart: The Skill-Policy Co-evolution Trap in Self-Improving Agents14m◆
News/Poisoned Identifiers Survive LLM Deobfuscation: A Case Study on Claude Opus 4.6
arxiv
PublishedApril 7, 2026 at 4:00 AM

Poisoned Identifiers Survive LLM Deobfuscation: A Case Study on Claude Opus 4.6

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

arXiv:2604.04289v1 Announce Type: cross Abstract: When an LLM deobfuscates JavaScript, can poisoned identifier names in the string table survive into the model's reconstructed code, even when the model demonstrably understands the correct semantics? Using Claude Opus 4.6 across 192 inference runs on

Stay posted· Newsletter

Three short emails a week: Monday, Wednesday and Friday. Top movers, new models and the most-referenced story.

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

Your email is used only to send The Bubble Brief. Privacy policy

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
arxivBridging the Gap: Converting Read Text to Conversational Dialogue14marxivTruth for Believable AI: Expressed Doubt, Provenance, and Belief Revision as an Engineerable Stance14marxivRICE-Alpha: Reliability-Informed Correction with Event Graphs for LLM-Agent Stock Forecasting14marxivThe World Won't Wait: Evaluating and Teaching Agents Under a Live Clock14m
The Bubble Brief
MON · WED · FRI

Read AI insights every Monday, Wednesday and Friday — top movers, new models, story of the week.

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

Your email is used only to send The Bubble Brief. Privacy policy

Originally published on arxiv ↗
PrivacyTermsRefundsLegal notice|AboutMethodologyContact

© 2026 Databubble

Built by Marouane Gazouzi · Readyo, AI interview coach

HomeModelsNews