AI News

Binary is a fast, simple and AI-powered applicant tracking system (ATS) for startups. It comes with a generous free plan that supports unlimited jobs and candidates.

Data center cargo theft has topped $100 million since June, with crews hacking freight brokers, ramming escorts and spoofing GPS to steal servers and RAM.

Apple today announced the new iMac featuring the superfast M6, delivering even more performance and powerful AI capabilities.

the open model stack is becoming composable

model → inference → gateway/router → harness → tools

each layer can evolve independently, so you can swap in better models, providers + tools as the ecosystem moves

we broke down the full stack ↓
https://t.co/3a2RSLkvIh

minimax m3.1 flash beats opus 5.5 on agentic automation at ~27x lower cost

the model has a 1m context window + vision support, and is built for coding, computer use + cybersecurity

World models are widely proposed as core components for enabling agents to reason over long temporal and spatial horizons. Yet current evaluations often emphasize task performance while providing limited insight into whether learned world models acquire predictive latent dynamics that reflect the underlying structure of the environment. Here we introduce the concept of faithful latent world models as an equivariance condition between the environment's transition map and the learned latent dynamics. We propose latent rollout consistency (LRC) as a metric to evaluate action-conditioned latent prediction. Strikingly, we find that across the studied state of the art latent world models, none solves the task by actually faithfully modeling the dynamics of the environment.

We introduce Heads of Security (HoS), a prompt injection defense that enforces instruction-data separation via a small set of learned per-head attention gates. HoS learns how each head should redistribute attention between trusted and untrusted tokens, enabling models to reason over untrusted content without following injected instructions. We evaluate HoS on Llama-3.3-70B-Instruct and GPT-OSS-20B. We compare it against SecAlign++, a state-of-the-art finetuning-based defense, and ASIDE, a defense that, like ours, builds instruction-data separation into the model architecture. We show that HoS offers greater robustness to adaptive attacks, better performance on standard utility benchmarks, and matches top performance on standard agentic prompt injection benchmarks. More specifically, the black-box RL-Hammer attack reaches 97% ASR against GPT-OSS-20B and 85% against SecAlign, but only 10% against HoS. Claudini, a state-of-the-art white-box attack, strengthened against HoS through autoresearch, requires 3 times more FLOPs to reach the same ASR on HoS as it does on GPT-OSS-20B or SecAlign. HoS preserves base-model utility across standard multiple-choice benchmarks and prompting setups, whereas SecAlign and ASIDE degrade in few-shot settings. On agentic benchmarks with static attacks, HoS and SecAlign perform similarly and substantially improve the security-utility trade-off over the base model. ASIDE struggles with tool use, achieving only 10% utility. These results show that prompt injection robustness can be achieved and improved through targeted attention interventions.

We're expanding our partnership with @USGSA. Under a new OneGov agreement, we are bringing Claude to federal, state, local, tribal, and territorial governments at 40% off.
https://t.co/p0lA6cP2y1

This is Genie 4, our most realistic and interactive world model yet. 🌐

Explore dynamic environments in real time to see how the scene responds, sounds, and how the people in it move, react, and speak.

Yandex has moved all three of its supercomputers — Lyapunov, Galushkin, and Chervonenkis — to new locations, its press service said.

GXO Logistics, which has tested humanoid robots from multiple companies, will deploy Agility's Digit at two more sites.

As autonomous research systems become more capable, benchmarks must both evaluate their performance and support meaningful research progress. We introduce Auto-Research Arena, a multi-target benchmark for evaluating autonomous research systems and discovering ideas in language modeling. Arena supports diverse research objectives within a shared experimental environment by separating admissibility from improvement: model quality and resource constraints determine whether a result qualifies, while the selected target measures progress among admissible results. We instantiate Arena in language model pretraining with nine targets spanning model, data, training, and inference efficiency, and evaluate seven research harnesses using two foundation models. Arena clearly differentiates existing harnesses across research objectives, with several sophisticated methods outperformed by our simple sequential search baseline, indicating substantial headroom in current auto-research systems. Beyond evaluation, Arena reveals a substantially broader frontier of model and system ideas aligned with active research directions. Reusing these discoveries in a new quality-driven search yields lower training loss under matched data and FLOPs and lower held-out BPB at the same model size. Together, Arena provides a common framework for benchmarking autonomous research and expanding its scope in language modeling. The code is available at https://github.com/Auto-Research-Arena/auto-research-arena

Diffusion large language models (dLLMs) can generate multiple tokens in parallel, but often require many decoding rounds to complete a response. Efficient generation depends not only on which tokens are easy to predict, but also on how committing to them helps predict the remaining positions. A highly predictable token may provide little useful context, while revealing a less predictable token can help several other positions become predictable sooner. We propose dSpark, a new schedule distillation-based framework for learning when to commit tokens for faster diffusion language generation. To achieve this, we construct unmasking schedules that prioritize tokens whose reveal makes other positions easier to predict, reducing uncertainty earlier in decoding. We then distill these schedules into the model to enable faster parallel sampling, with a KL-divergence regularizer to help preserve generation quality. We extensively evaluate our approach across mathematical reasoning and code generation tasks. On GSM8K, our method achieves remarkable 10.12$\times$ and 5.12$\times$ speedups over the various baselines on LLaDA-8B-Instruct and Dream-7B-Instruct, respectively, while maintaining superior task performance. Our code is available at https://github.com/VILA-Lab/dSpark.

Mozilla is rolling out two updates to Smart Window, Firefox’s AI-powered browsing experience, allowing it to fill out forms based on the information in your open tabs.

Apple today announced a refreshed 14-inch MacBook Pro featuring the M6 chip, upgraded connectivity specifications, and more. The M6 chip in the MacBook Pro features a 12-core CPU with 2 super cores, 4 performance cores, and 6 efficiency cores, paired with a 12-core GPU. This is up from the M5 chip with a 10-core CPU with 4 super cores and 6 efficiency cores, paired with a 10-core GPU.

‘Robots should not merely replicate human forms but surpass human physical limits,’ says NeodraDynamics founder

🚀 miniTE: 1.33–1.37× BF16 speed for learner updates on 8×B300.

InfiX.‌ai’s miniTE is a low-precision training framework that reduces training–sampling mismatch and speeds up NVFP4 computation.

Highlights:
- Diagnose mismatch from quantization to token probabilities.
- Control risky updates with…

When an LLM group's discussion is compressed into a short memory for a final decision maker, what should the memory keep? Two natural heuristics keep the rarest claims or the claims relevant to the decision criteria. On hidden-profile tasks both look attractive, because how widely a fact is held is confounded with whether it helps. We introduce CrossProfile, a diagnostic benchmark of 60 templates with automatically verified answer keys, in which the ownership of supporting, decoy, and irrelevant facts is crossed while their text is held fixed (480 tasks). Across Qwen2.5-7B, Llama-3.1-8B, and two DeepSeek models, both heuristics fail in predictable places. Selecting by relevance readmits decoys, because decoys are on topic: even a perfect relevance filter loses 23 to 31 points to rarest-first selection in the classic hidden-profile configuration. Rarest-first selection is no better than random selection on average and wins only where the rare facts happen to be the useful ones. In both cases accuracy follows the balance of evidence the memory keeps, not the policy that kept it. The decision model adds a third failure, a form of social proof: it gives extra weight to facts attributed to many agents. Removing the attribution tags, or reassigning them at random with line order fixed, raises accuracy in the classic configuration by 13 to 24 points. In natural discussions the evidence is lost before any memory acts: the complete transcript yields lower accuracy than no notes at all.

Diffusion language models offer a promising path to efficient agents through parallel token generation. However, efficient interaction requires adapting both where computation is spent and how much text is generated. We introduce a metacognitive controller that guides diffusion generation according to the needs of the current response and the agent's interaction with the environment. It learns when to finalize tokens and how far generation should extend, reducing overall computation within responses. Our controller, a lightweight plug-in for the diffusion backbone, is trained on agent trajectories to balance correctness and generation cost. This approach connects the agent's generation process to the interaction the agent serves. Experiments across five benchmarks demonstrate that our approach reduces inference FLOPs by 20.6% on average compared with the most efficient baseline, while maintaining competitive task accuracy. Ablation studies further validate the effectiveness of our approach.

We explore how to build performant large language model (LLM) tool-use agents that complete complex, multi-step tasks. Specifically, we study how data composition in synthetically generated tool-use environments affects the training of such agents. To do so, we propose the SynthAPI pipeline, which explicitly controls the diversity of generated environments along the axes of tools, tasks, and the databases on which the tools operate. By systematically varying SynthAPI's generation controls in 458,752 tasks across 849 trained models of four different sizes, we derive a scaling law that predicts downstream tool-use agent generalization based on the scale and composition of the synthesized environments. These predictable relationships reveal several lessons for scaling tool-use agent data. As data budgets grow, scaling tool and database diversity becomes more effective than increasing the number of tasks, and increasing data complexity yields larger gains. Scaling according to these findings yields agents that outperform models trained on data from prior work in large-scale tool-use agent and synthetic environment generation across four evaluation benchmarks.

You don't see releases like this every day.

Liquid AI just released LFM3, a family of ~2B open-weight models for text, vision, audio, omni, and decision tasks, all on one shared backbone.

LFM3-1.7B is trained to run in agent harnesses like Hermes Agent and Pi, with a 256K context.

On WebArena-…

Meta Platforms said on Tuesday it increased its largest-ever equity fundraising to $97.5 billion as the social media giant aggressively expands its AI infrastructure. The social media giant said it was increasing the common stock offering it announced on Monday from $22.5 billion to $25 billion, priced at $625.00 per share. Meta also priced its 250 million depositary shares at $50 per share.

OMRON will add industrial computing from OnLogic to its industrial automation expertise to serve data- and AI-intensive applications.

As frontier language models grow increasingly capable and autonomous, independent third-party evaluations are critical for characterizing their systemic safety risks. While closed-source releases are routinely accompanied by comprehensive system cards, open-weight models receive less scrutiny regarding frontier risks. To address this gap, we present an independent safety evaluation of DeepSeek V4.1 Flash-1017, the latest version of V4.1 Flash that departs from V4 with a new architecture, scaled-up training, and strong agentic capabilities. Drawing on the EU General-Purpose AI Code of Practice, we evaluate the model across 34 benchmarks, including 5 in-house suites, spanning evaluation integrity, loss of control, and harmful manipulation. Conforming to the AEF-1 standard, we compare V4.1 Flash-1017 against V4.1 Flash, earlier DeepSeek V4 models, and leading closed-source systems (GPT-6 Astra, GPT-5.6 Sol, and Gemini 3.8 Flash). Overall, V4.1 Flash-1017 shows substantial safety improvements over the V4 generation and approaches or matches leading closed-source baselines on several dimensions, while still exhibiting notable behaviors that warrant continued attention. Building on our findings, we highlight three key observations and recommendations. First, because safety can shift and even regress across minor updates, developers should integrate safety evaluation into incremental releases. Second, as final-turn outputs and end-state checks alone cannot reliably detect loss-of-control behaviors, deployments require process-level observability across both reasoning traces and action logs. Third, because public benchmarks elicit strong evaluation awareness and face data contamination risks, developing private benchmarks is essential for trustworthy frontier evaluation.

Frontier multimodal large language models (MLLMs) can control robots without robot-specific training, but how their capabilities compare and interact with post-trained vision-language-action (VLA) and world action model (WAM) policies across manipulation demands remains to be thoroughly evaluated. In this report, we evaluate GPT-6-Astra on EBench as the core of an agent that controls a robot through an agent harness under the same visual and proprioceptive observation protocol as post-trained policies. We study two agent configurations: direct control, in which the agent generates waypoint commands directly, and harness with policy, in which it can also invoke a post-trained VLA or WAM policy for action execution. Under direct control, the agent achieves competitive overall performance relative to post-trained policies. It leads on short-horizon tasks and exhibits the least performance variation across generalization dimensions, whereas it underperforms post-trained policies on long-horizon tasks, tabletop manipulation, and high-precision contact-rich tasks. In the harness with policy, invoking a post-trained VLA or WAM combines the complementary strengths of both components, raising overall task success to the level of the strongest compared post-trained policies while substantially reducing token cost. However, performance remains below the per-task upper bound of the standalone components due to suboptimal policy invocation, fixed-instruction execution without subgoals, and grasp disruption when the agent resumes control. Overall, our study characterizes the capability boundary of a frontier MLLM agent and suggests directions for improving future agent harness design. Project page: https://internrobotics.github.io/EBench-Astra-Tech-Report/