arXiv:2609.01345v2 Announce Type: replace Abstract: Inference cascades cut cost by answering most queries with a cheap model and escalating a hard tail to…
AI News Brief Hourly Summary 2026-09-08 05h : 12 posts
12 posts published in the last hour 02:32Semantic Overlays: Mitigating Prompt Injection with Annotations Beyond Tokens and Steering Vectors 02:32FORESIGHT-9: Prospective and Process-Aware Evaluation of Adaptive Trading Agents 02:32AI Revealed Preferences 02:32$A^2E$ : An End-to-End Agent Auditing Engine 02:32NxN E-valuation:…
Semantic Overlays: Mitigating Prompt Injection with Annotations Beyond Tokens and Steering Vectors
arXiv:2608.23873v3 Announce Type: replace Abstract: Everything a language model sees is tokens. The serving stack knows what each span is — user input,…
FORESIGHT-9: Prospective and Process-Aware Evaluation of Adaptive Trading Agents
arXiv:2608.29372v2 Announce Type: replace Abstract: Retrospective backtests provide a limited test of adaptive trading agents: they cannot rule out…
AI Revealed Preferences
arXiv:2608.26178v2 Announce Type: replace Abstract: There is growing interest in whether language models have stable preferences, for technical, safety,…
$A^2E$ : An End-to-End Agent Auditing Engine
arXiv:2608.07346v3 Announce Type: replace Abstract: With the rapid advancement of large language models (LLMs), harnesses have become essential…
NxN E-valuation: Hypothesis Certification via a Conformal CRT Null
arXiv:2608.06621v3 Announce Type: replace Abstract: We propose NxN E-valuation, a handy, e-value-based hypothesis-certification algorithm that lets a…
KernelGenBench: A Multi-Source and Multi-Chip Benchmark for LLM-based Kernel Generation
arXiv:2607.27231v2 Announce Type: replace Abstract: Modern AI systems depend on specialized accelerator kernels, whose development is complicated by…
EvoCUA-1.5: Online Reinforcement Learning for Multi-turn Computer-Use Agents
arXiv:2607.09773v2 Announce Type: replace Abstract: Computer-use agents must solve long-horizon tasks through repeated interaction with partially…
Aletheia: An Offline-First Clinical Decision Support System for Differential Diagnosis in Low-Resource Healthcare Settings
arXiv:2607.24814v3 Announce Type: replace Abstract: Access to specialist clinical expertise remains severely limited across sub-Saharan Africa, where…
Improving Weak World Models Behind Strong Agents in Atari Pong
arXiv:2607.15142v3 Announce Type: replace Abstract: Strong world-model agents frequently contain weak world models. We study this agent-world-model gap by…
Firmus Signs OpenAI as Anchor Customer for Malaysian AI Factories
Australian AI infrastructure company Firmus Technologies and OpenAI announced a multi-year strategic partnership on September 8, 2026, under which OpenAI…
GPTNT: Benchmarking Real-Time Collaboration Between Multimodal Agents on Keep Talking And Nobody Explodes
arXiv:2606.28514v2 Announce Type: replace Abstract: Multimodal models are increasingly deployed to solve tasks collaboratively with humans or other…
AI News Brief Hourly Summary 2026-09-08 04h : 12 posts
12 posts published in the last hour 01:32Robust and Efficient Guardrails with Latent Reasoning 01:32BUZZY: Contrastive Scoring to Mitigate Text-Induced Bias in Multimodal Multiple-Choice QA 01:32Role-Aware Artificial Intelligence Across Augmentation and Automation in Human-Machine Symbiosis 01:32Decision-Aware Memory Cards: Counterfactual-Inspired Context…
Robust and Efficient Guardrails with Latent Reasoning
arXiv:2605.29068v2 Announce Type: replace Abstract: Maintaining the safety of large language models (LLMs) is crucial as they are increasingly deployed in…
BUZZY: Contrastive Scoring to Mitigate Text-Induced Bias in Multimodal Multiple-Choice QA
arXiv:2603.28026v4 Announce Type: replace Abstract: Multimodal multiple-choice question answering (MCQA) provides a standardized and objectively…
Role-Aware Artificial Intelligence Across Augmentation and Automation in Human-Machine Symbiosis
arXiv:2605.00440v2 Announce Type: replace Abstract: The evolution of artificial intelligence (AI) has rendered the boundary between humanity and…
Decision-Aware Memory Cards: Counterfactual-Inspired Context Selection and Compression for Tool-Using LLM Agents
arXiv:2606.08151v3 Announce Type: replace Abstract: Modern large language model (LLM) agents do not simply need longer contexts; they need…