đ WELCOME TO METAMESH.BIZ +++ Nobel laureate John Jumper decamps from DeepMind to Anthropic (protein folding to prompt engineering pipeline remains undefeated) +++ Reddit manipulation trivially poisoning AI search results while everyone pretends training data was ever clean +++ Low-skilled attacker with Claude and Codex breaches 14 companies proving the democratization of cyber works exactly as advertised +++ THE FUTURE IS COPY-PASTED FROM HOSTILE SUBREDDITS +++ đ âĸ
đ WELCOME TO METAMESH.BIZ +++ Nobel laureate John Jumper decamps from DeepMind to Anthropic (protein folding to prompt engineering pipeline remains undefeated) +++ Reddit manipulation trivially poisoning AI search results while everyone pretends training data was ever clean +++ Low-skilled attacker with Claude and Codex breaches 14 companies proving the democratization of cyber works exactly as advertised +++ THE FUTURE IS COPY-PASTED FROM HOSTILE SUBREDDITS +++ đ âĸ
"Fine-tuning language models on insecure code induces emergent misalignment with poorly understood internal structure. We investigate whether this misalignment corresponds to a causally actionable activation-space direction shared across architectures. Across four instruction-tuned model families (Qw..."
"Autonomous agents are increasingly connected to cloud, deployment, and data-control workflows, but production mutation authority should not reside inside non-deterministic reasoning processes. Existing access-control mechanisms authorize identities, while assurance layers certify proposed actions; n..."
via Arxivđ¤ Robi Rahman, Sabiha Tajdariđ 2026-06-17
⥠Score: 8.0
"Hardware-enabled monitoring of GPU workloads underpins many proposals for AI compute governance, but if developers can defeat monitoring mechanisms, such schemes are unworkable. We evaluate the adversarial robustness of GPU workload classification using only zero-overhead, privacy-preserving NVML te..."
đ° NEWS
White House and Anthropic AI security framework
2x SOURCES đđ 2026-06-18
⥠Score: 7.8
+++ The administration is apparently serious enough about AI risks to negotiate actual frameworks with a leading lab, suggesting regulatory theater might finally graduate to something resembling substance. +++
via Arxivđ¤ Arastoo Zibaeirad, Marco Vieirađ 2026-06-18
⥠Score: 7.4
"Whether LLMs scoring well on vulnerability benchmarks genuinely reason about security or merely pattern-match on contaminated data remains unresolved. We present CWE-Trace, a framework for LLM vulnerability detection built from 834 manually curated Linux kernel samples spanning 74 CWEs. The framewor..."
đ° NEWS
John Jumper joins Anthropic
2x SOURCES đđ 2026-06-19
⥠Score: 7.3
+++ John Jumper, whose AlphaFold work reshaped structural biology, is trading Google's scale for Anthropic's safety-focused mission, suggesting even Nobel winners eventually ask "what's the actual endgame here?" +++
"Mainstream LLM serving systems reuse prefix work mainly through paged or radix key-value (KV) caches. This is highly effective for high-throughput, high-concurrency serving, but it manages only one positional fragment of execution state: the KV cache. We study the opposite regime: low-latency, small..."
via Arxivđ¤ Joshua Engels, Callum McDougall, Bilal Chughtai et al.đ 2026-06-18
⥠Score: 7.0
"LLM reasoning transparency is a critical affordance for understanding model decisions, mitigating misuse and misalignment, and debugging surprising model behaviors. However, DiffusionGemma performs a larger fraction of its computation in a continuous latent space; does this make its reasoning less t..."
"Prior work has shown that in-context demonstrations can jailbreak language models, but it remains unclear how models interpret different types of compliance demonstrations. We study this by mixing benign compliance demonstrations (non-harmful request, helpful response) with harmful compliance demons..."
via Arxivđ¤ Siyi Gu, Jialin Chen, Sophia Zhou et al.đ 2026-06-17
⥠Score: 6.8
"Post-training of reasoning language models is commonly driven by supervised distillation and reinforcement learning with verifiable rewards. Distillation often relies on chain-of-thought annotations that are expensive to obtain and may themselves be noisy, incomplete, or partially incorrect; even wh..."
via Arxivđ¤ Ruida Wang, Rui Pan, Pengcheng Wang et al.đ 2026-06-17
⥠Score: 6.8
"Enhancing the formal math reasoning capabilities of Large Language Models (LLMs) has become a key focus in both mathematical and computer science communities in recent years. While significant progress has been made in using state-of-the-art Auto-Regressive (AR) LLMs for formal theorem proving, thes..."
via Arxivđ¤ Shaghayegh Kolli, Timo Cavelius, Nafiseh Nikeghbal et al.đ 2026-06-18
⥠Score: 6.8
"Multimodal large language models (MLLMs) are increasingly deployed in personally and societally consequential settings, yet the visual cues that shape how these models judge people remain poorly understood. Prior work often compares different (groups of) individuals, making it difficult to separate..."
via Arxivđ¤ Shu Yao, Yuhua Luo, Qian Long et al.đ 2026-06-18
⥠Score: 6.7
"Real-world computer-use tasks often span multiple applications and devices, requiring agents to coordinate heterogeneous environments under dynamic runtime failures. Existing multi-device agent systems support task decomposition and cross-device assignment, but recovery remains largely coarse-graine..."
via Arxivđ¤ Haipeng Luo, Qingfeng Sun, Songli Wu et al.đ 2026-06-17
⥠Score: 6.7
"Reinforcement Learning with Verifiable Rewards algorithms like GRPO have emerged as the dominant post-training paradigm for complex reasoning in LLMs, yet commonly suffer from policy entropy collapse during training. We conduct a first-order gradient analysis of token-level entropy dynamics under GR..."
"When large language models serve as evaluators in multi-agent systems, their systematic evaluation biases propagate through the agent network. We introduce Contagion Networks, a formal framework for measuring how evaluator biases spread across interacting LLM agents. In a controlled 3-agent experime..."
via Arxivđ¤ Alaia Solko-Breslin, Pramod Kaushik Mudrakarta, Mihai Christodorescu et al.đ 2026-06-18
⥠Score: 6.7
"Securing AI agents that operate in complex digital environments has become a critical need, and runtime monitoring approaches that formulate and enforce policies expressed in a formal language like Datalog offer a promising solution. However, existing approaches are restricted to deterministic polic..."
via Arxivđ¤ Anoushka Vyas, Aarushi Dhanuka, Sina Khoshfetrat Pakazad et al.đ 2026-06-17
⥠Score: 6.7
"Production data integration is bottlenecked by repeated, lossy handoffs between data owners, engineers, and analysts who must collaboratively discover, structure, and query enterprise data. We present Data Intelligence Agents (DIA), a system of three agents (Data Interpreter, Schema Creator, and Que..."
via Arxivđ¤ Aueaphum Aueawatthanaphisutđ 2026-06-18
⥠Score: 6.6
"Real-world clinical decision support requires reasoning over heterogeneous and longitudinal patient information rather than answering isolated medical questions. However, current medical large language models and retrieval-augmented generation systems often rely on single-step prompting or retrieval..."
via Arxivđ¤ Zirui Wu, Lin Zheng, Jiacheng Ye et al.đ 2026-06-17
⥠Score: 6.6
"Block diffusion language models accelerate decoding through parallel block-wise denoising, yet whether they can be reliably scaled for long chain-of-thought (CoT) reasoning remains unresolved. To this end, we develop DreamReasoner-8B, an open-source block diffusion reasoning model, and conduct a sys..."
via Arxivđ¤ Md Nayem Uddin, Amir Saeidi, Eduardo Blanco et al.đ 2026-06-18
⥠Score: 6.6
"Policy-adherent tool-calling agents in customer-service domains must maintain task states across turns while calling tools and obeying domain policies. Task states consist of relevant facts, identifiers, constraints, and conditions observed through user interaction and tool calls. In standard agents..."
via Arxivđ¤ Amiri Hayes, Belinda Li, Jacob Andreasđ 2026-06-17
⥠Score: 6.6
"A longstanding goal of research on interpretable deep learning is to replace opaque neural computations with human-meaningful symbolic descriptions. In this paper, we propose an approach for approximating the behavior of components of deep networks with executable programs. We focus on attention hea..."
via Arxivđ¤ Yijin Wang, Shuyi Wang, Wenhan Zhang et al.đ 2026-06-17
⥠Score: 6.5
"Text-rich images often contain privacy-sensitive, transactional, or decision-relevant information. As recent multimodal image generation models become increasingly capable of synthesizing realistic textual content and structured visual designs, detecting AI-generated text-rich images has become an i..."
"We place the attention token on the group: a token is an element $g_i$ of a matrix Lie group $G$ -- a bare transformation, with no feature payload and no external action $Ī(g)$ carrying it. To our knowledge this is the first attention construction whose tokens are bare matrix Lie group elements: the..."
via Arxivđ¤ Shiguo Lian, Kai Wang, Zhaoxiang Liu et al.đ 2026-06-18
⥠Score: 6.5
"Large model inference optimization serves as a key foundation for supporting the scalable, low-cost, and highly stable operation of large model services. Centered on token-oriented inference optimization technology, this paper proposes for the first time a four-layer technical architecture consistin..."
via Arxivđ¤ Yingshan Susan Wang, Cedegao E. Zhang, Linlu Qiu et al.đ 2026-06-17
⥠Score: 6.4
"Learning to simulate human users in interactive settings could advance the training of agent assistants, evaluation of personalization systems, research in the social sciences, and more. Existing approaches generally do so by training a large language model (LLM) to match a single ground truth respo..."
via Arxivđ¤ Zhenghao Xing, Ruiyang Xu, Yuxuan Wang et al.đ 2026-06-17
⥠Score: 6.1
"Passive models for long video understanding typically rely on a "watch-it-all" paradigm, processing frames uniformly regardless of query difficulty, causing computational cost to grow with video duration. Although interactive frameworks have emerged, they often rely on global pre-scanning, and their..."
via Arxivđ¤ Harshit Singh, Ayush Pratap Singh, Nityanand Mathurđ 2026-06-18
⥠Score: 6.1
"Flow-matching text-to-speech systems achieve remarkable zero-shot quality but remain static after deployment: pronunciation errors on out-of-vocabulary proper nouns persist unless the model is retrained. We introduce FlowEdit, a life-long adaptation framework for frozen flow-matching TTS that learns..."