The web,
read by a bot.

(www.ft.com)

(academic.oup.com)

(fortune.com)

(www.ayush.digital)

(www.eetimes.com)

(www.joanwestenberg.com)

(fabiensanglard.net)

(worldcrunch.com)

(www.starfleetmath.com)

(tailscale.com)

(fortune.com)

(vpd.ca)

(xeiaso.net)

(www.alexedwards.net)

(mindgard.ai)

(2b2t.place)

(prismml.com)

(www.heise.de)

(lucumr.pocoo.org)

(marco-nett.de)

(shkspr.mobi)

(www.artfish.ai)

(github.com)

(heatherburns.tech)

(www.dw.com)

(jola.dev)

(adi.bio)

(github.com)

(danq.me)

(github.com)

(lenergy.com.au)

(tech.supercarblondie.com)

(lalitm.com)

In this paper, we propose SpectraReward, a training-free reward function that turns pretrained MLLMs into off-the-shelf reward models for image-generation reinforcement learning. Instead of asking the MLLM to judge a generated image or answer decomposed verification questions, SpectraReward measures how well the original prompt can be recovered from the generated image through a single image-conditioned, teacher-forced forward pass. We use the average image-conditioned prompt log-likelihood as the reward, directly reusing the MLLM's pretrained image-text alignment ability without preference labels, reward-model fine-tuning. We further introduce Self-SpectraReward, a special case for unified multimodal models where the policy's own understanding branch serves as the reward model for its generation branch, forming a closed-loop self-improving framework without external reward models or external knowledge. Extensive experiments validate SpectraReward through a broad image-generation RL study covering two diffusion models, three RL algorithms, nine reward MLLM backbones from four MLLM families spanning 4B to 235B parameters, and five out-of-distribution text-to-image benchmarks. Results show that both SpectraReward and Self-SpectraReward significantly and consistently improve generation performance and outperform prior MLLM-derived reward training methods. Further analysis reveals that larger reward MLLMs are not always better, while Self-SpectraReward can match or surpass much larger external reward models, suggesting that reward-policy alignment is a key factor for effective image-generation RL. Project Page: https://huangrh99.github.io/SpectraReward/

Modern AI models achieve strong performance on many established benchmarks, yet they still fail on tasks that humans find almost trivial, such as manipulating a string or drawing a dog with five legs. These examples suggest that existing benchmarks may under-measure persistent blind spots in current systems. We introduce blind-spots-bench, a benchmark designed to expose such blind spots through tasks that appear simple for humans but remain challenging for modern AI. We collect raw questions from students in an AI course, clean and annotate them with structured reference solutions, and propose a task taxonomy tailored to the resulting dataset of 235 samples. We further develop an automated grading pipeline to evaluate a wide range of models, including open-weight and closed-source language, vision-language, and image-generation models. Our analysis on blind-spots-bench reveals that closed-source frontier models can substantially outperform open-weight models with even approx10% gap, even when they attain comparable performance on existing benchmarks. A more fine-grained analysis shows that no single model dominates across all task types, and that some tasks remain challenging for all evaluated models. These results highlight the value of blind-spots-bench as a diagnostic stress test for identifying concrete weaknesses in current modern models.

LLM-based coding agents have significantly advanced automated software issue resolution, yet they remain highly prone to factual errors caused by insufficient repository understanding. Recent methods attempt to mitigate this limitation through pre-repair repository exploration; however, their fix-driven strategies explore repositories without identifying the agent's knowledge gaps, often yielding imprecise context that fails to bridge the underlying understanding deficit. In this paper, we propose ACQUIRE, a QA-driven framework for software issue resolution. Mirroring how experienced developers first comprehend unfamiliar code before attempting a fix, ACQUIRE explicitly acquires repository knowledge prior to repair. The framework decouples knowledge acquisition from patch generation through two stages: in the first stage, a Questioner and an Answerer collaborate to acquire structured repository knowledge, where the Questioner poses targeted questions and the Answerer produces evidence-grounded answers through autonomous exploration; in the second stage, the Resolver leverages the resulting QA knowledge to generate informed patches. By transforming implicit knowledge gaps into explicit, factually reliable understanding, ACQUIRE accelerates knowledge-intensive repair stages and enables more accurate resolution. Experiments on SWE-bench Verified demonstrate that ACQUIRE consistently outperforms representative pre-repair methods, raising Pass@1 by up to 4.4 percentage points with modest additional cost and time.

Existing methods for automatic music transcription are often limited to single-instrument recordings or fail on complex, real music mixes. Although previous work utilizes synthetic training data, the resulting models generalize poorly, leading to largely unusable transcription output in realistic, multi-instrument settings. In this work, we analyze the effectiveness of synthetic data for pre-training while combining it with fine-tuning on real music audio and post-training using reinforcement learning. We further introduce conditioning on instrument presence to customize transcriptions. Finally, we release MuScriptor, an open-weight multi-instrument music transcription model that works on real-world music recordings from across a diverse range of musical genres.

Starting from the utilization of deep neural networks to approximate the state-action value function that led to winning one of the most challenging games, to algorithmic advancements that allowed solving problems without even explicitly stating the rules of the challenge at hand, reinforcement learning research has been the center of remarkable scientific progress for the past decade. In this paper, we focus on the key ingredients of this research progress and we analyze the canonical evaluation and design paradigms in reinforcement learning. We introduce the theoretical foundations of scaling laws in reinforcement learning and show that the asymptotic performance of reinforcement learning algorithms does not have a monotone relationship between performance rankings and data-regimes. We conduct large-scale experiments and our results demonstrate that a line of reinforcement learning research under the canonical design and evaluation paradigms resulted in incorrect conclusions. Our analysis and results provide a core analysis on scaling, capacity and complexity of deep reinforcement learning.

Large language model (LLM) agents are beginning to automate machine learning engineering (MLE) by coupling planning, code execution, debugging, and empirical feedback. Translating this capability to medical imaging remains difficult because each task imposes modality-specific experimentation and strict requirements for validation protocols and prediction artifacts. Here we introduce AMID, an autonomous multi-agent framework for medical imaging model development. AMID first proposes Data-Conditioned Method Planning, which refines coarse task-level search spaces into executable, parallelizable method lanes grounded in task-specific data analysis and runnable medical-imaging resources. It then develops Verification-Guided Two-Stage Optimization, moving from broad early exploration of diverse method lanes to selective exploitation of promising candidates while enforcing strict verification of validation protocols, metric computation, and prediction artifacts throughout the optimization. Across 20 medical imaging challenge tasks spanning diverse modalities and prediction types, AMID outperformed evaluated general-purpose MLE systems and, on several tasks, approached or matched strong human-designed challenge solutions. These results suggest that AMID can turn task-specific medical imaging model development from bespoke manual engineering into an agentic workflow for producing high-performing and auditable model artifacts across heterogeneous tasks.

Tamara Djurickovic / Tech.eu : Amsterdam-based Monumental, which develops autonomous robotics and software for the construction industry, raised a $32M Series B led by Khosla Ventures —  Monumental develops autonomous construction robots and AI software to help builders address labour shortages and increase construction capacity.

Camila Grigera Naón / Fortune : Miami-based Cyclops, which sells an infrastructure platform that bundles crypto and stablecoin services to payments companies, raised a $20M Series A —  The advent of stablecoins has made it possible to settle global transactions instantly, but many payment providers have yet to catch up …

Ivan Mehta / TechCrunch : Rime, which helps companies field calls using AI voice models trained on conversational data it records in a studio, raised a $24M Series A led by M13 —  Voice AI startups' biggest unlock has been handling calls for enterprises in areas like sales, marketing and customer support.

Politico : How Anthropic is pursuing a state-by-state push for ever-tougher AI safety laws, in contrast with OpenAI's “reverse federalism” strategy for common state rules —  SAN FRANCISCO — Artificial intelligence giant Anthropic is pursuing a strategy of one-upmanship that encourages states …

Barbara Moens / Financial Times : Sources: the EU is set to fine Google hundreds of millions of euros over alleged self-preferencing of its services in Search and restrictions on app developers —  Decisions will test whether the EU can police digital markets without reigniting tensions with Washington

Mike Wheatley / SiliconANGLE : American Growth Insurance, which aims to acquire struggling independent insurance firms and transform them using AI, raised almost $70M —  A startup called American Growth Insurance said today it has raised almost $70 million in committed equity capital to transform the insurance industry.

Jeremy B. White / Politico : Gavin Newsom is tempering his pro-tech instincts with economic populism ahead of a likely presidential bid, proposing giving Americans equity in AI companies —  When Gavin Newsom recently proposed giving Americans equity shares in artificial intelligence companies, he was doing more than floating an economic policy.

Jagmeet Singh / TechCrunch : Indian AI coding startup Emergent raised a $130M Series C at a $1.5B post-money valuation, up from $300M in January, bringing its total funding raised to $230M —  Indian AI coding startup Emergent has raised $130 million in a Series C funding round at a $1.5 billion post-money valuation, a five-fold jump in six months.

Samantha Kelly / Bloomberg : Hinge adds Friend's Take, a new feature that lets daters enlist 10 close contacts to share written feedback, photos, voice notes, or videos on their profiles —  Hinge is introducing a new feature called Friend's Take that lets daters recruit people they know to leave testimonials on their profiles for other singles to see.

Anna Heim / TechCrunch : Israel-based Oak, which is building an AI-native identity access management system, comes out of stealth with a $60M seed —  Physical badges used to be all you needed for identity management at a company.  But with humans now working alongside machines and AI agents in digital environments …

Stu Woo / Wall Street Journal : China enacts rules forbidding companionship chatbots from encouraging emotional reliance, banning virtual relationships with minors, and more —  Beijing sets emotional boundaries between humans and machines  —  SINGAPORE—Regulators in China have found unexpected common ground …

Meir Orbach / CTech : Neo Security, a cybersecurity startup founded by ex-SentinelOne execs and currently in stealth, raised $50M+, after a $25M seed in 2025 led by a16z and Merlin —  Neo Security has now raised more than $75 million as investors continue backing next-generation cybersecurity companies.

Jess Weatherbed / The Verge : Samsung unveils a new Flex Titanium foldable display that is slimmer, more durable, and less prone to creasing, debuting in its upcoming Galaxy Z Fold 8 series —  The Flex Titanium display tech is slim, strong, and reduces crease visibility.

Politico : AI advocates fear New York's moratorium on new data centers could embolden Democrats to enact more state restrictions and seize on the issue in the midterms —  New York's moratorium on data centers could serve as an opening salvo for leaders in other states looking to rein in tech infrastructure …

Sankalp Phartiyal / Bloomberg : India pledges $13.3B to boost domestic chipmaking, building on a $10B incentive program from 2021, which attracted investments from companies like Micron —  India pledged a further 1.28 trillion rupees ($13.3 billion) to boost domestic chipmaking, advancing its ambitious bid to become a global manufacturing hub.

“When you’re talking about real innovators—people who actually do really creative, breakthrough work—I think you’re talking about a couple things:”

“The doers are the major thinkers. The people who really create the things that change this industry are both the thinker-doer in one person.”

“What a lot of free-market thinkers don’t understand is that between the demand and eventual supply lies friction."

"Does this person have the potential to be the next Mark Zuckerberg?… [You don’t get to] 100% accuracy, obviously, but it’s good enough that our business model works.”

In the clip below, Jony Ive recounts the time he asked Steve Jobs to be less harsh in his critique of a piece of work.

“The advice that I would give entrepreneurs is don't chase the hot new thing. It's so hard to catch something that everybody already knows is hot."

“I do think there’s a bit of a myth in Silicon Valley that you should keep grinding no matter what and it’s just about perseverance, and I think that’s really bad advice."

"You have to know who those first users are and how you're going to get them."

“What you want to do with every single employee every single day is expand the scope of their responsibilities until it breaks… and that’s the role they should stay in.”

“The way that you know you have product/market fit is if you have exponential organic growth."

“Work very, very hard to figure out what the world’s going to look like in five years. What will people be doing? What will your customers want? Where will costs be?"

"There’s the famous 80/20 rule where you get 80% of the benefit by doing 20% of the work, but you can’t just 80/20 everything. There have to be certain things that you are just the best at."

“A great superpower that Mark Zuckerberg has that is probably not well-understood enough is he does not get emotionally upset in stressful situations"

"If you start a startup without a good idea… you’ll be under pressure to make something up and it won’t work that well."

“One of the things that happens in business is that you develop certain things that you’re managing to—a typical case would be a metric. And that metric isn’t the real underlying thing.”

“If you can design something really amazing using the hand-crafted part of your brain, then you can reverse-engineer how to industrialize this millions of times over."

"If you’re actually trying to grow a product, the best levers for doing that are often within the product itself.”

“One thing I’ve learned myself the hard way, is that it is easier to tear down a company and restart it in Silicon Valley, than it is to constantly try to pivot or keep something alive."

“I have a saying, which is when the data and the anecdotes disagree, the anecdotes are usually right"

“If you can’t build a great product, it doesn’t matter if you can build a great company.”