The Archive

ARMT enables long-context LLM processing with constant memory scaling via associative recurrent architecture.

Empirical audit reveals trained distributional RL agents' risk estimates often violate first-order stochastic dominance.

Formal language theory on minimal coloring schemes for identifiability in Gold's language-learning model.

Differentially private collaborative Bayesian optimization achieves centralized performance without raw-data sharing.

Interaction-based test-time compute (model-environment feedback loops) outperforms longer reasoning or sampling at inference.

Bangla hate-speech detectors fail on implicit, context-dependent cases; benchmark-trained models miss cultural nuance.

Google DeepMind and AIM launch ATL Saathi, a Gemini-powered educational tool for Indian robotics labs.

Experts explain how they work, what they can do, and what's still unsettled.

Apple sues OpenAI over alleged trade secret theft; Stratechery frames it as reactive rather than substantive legal action.

Waze is getting an AI makeover. Google is integrating its flagship AI assistant, Gemini, into the driving app with the goal of letting users personalize their trips a little more. Of the four new updates, only two are being described as involving Gemini. Waze says its updating its conversation reporting feature, first introduced in 2024, to allow drivers to use conversational voice commands to report traffic incidents and suggest map updates, like a road closure or outdated house number. In addition, Waze introduced Destination Search, enabling drivers to use (again) use conversation voice co...

Simon Willison examines DRI (Directly Responsible Individual) concept from Apple/GitLab in context of LLM agents, arguing humans must retain accountability.

shot-scraper 1.11 release adds JavaScript file loading and extends server startup timeout from 1s to 30s.

Anthropic extends Claude Fable 5 availability through July 19 on paid plans, citing compute constraints and GPT-5.6 Sol positioning.

sqlite-utils 4.1.1 adds TransactionError guard for table.transform() with foreign keys to prevent silent data loss.

Lorde performing at the 2026 Governors Ball. | Photo: Siegfried Anthony/Billboard via Getty Images Lorde was performing at the Real Cool Festival in Madrid on Thursday and took some time during her set to speak out against AI glasses. While she didn't specify any brands in particular, it's likely she was taking a shot at festival sponsor Ray-Ban, which has collaborated with Meta on a pair of AI smartglasses. The comments were captured in videos shared to social media. After thanking the crowd for being there and taking part in "something real," she said that it was increasingly hard to know i...

Apple's self-driving car program never really got off the ground, but it may have been what made the company's chips the powerful AI performers they are. Early in the development of the self-driving platform, Apple realized that it would need powerful on-device AI processing. While the car processor was never finished, as Mark Gurman details in his latest Power On newsletter, it did lead to the development of the Neural Engine, the backbone of Apple's on-device AI processing. The Neural Engine made its debut with the iPhone X and the A11 Bionic. In those early days, it was primarily used for ...

A yard sign opposing a planned data center is displayed along Route 54 in Mount Carmel Township Northumberland County. | Image: Getty Images This is The Stepback, a weekly newsletter breaking down one essential story from the tech world. For more on the data center buildout, follow Emma Roth. The Stepback arrives in our subscribers' inboxes on Sunday at 8AM ET. Opt in for The Stepback here. How it started Years before the AI boom threatened local power grids, a small group of protesters set the stage for the battles cropping up across communities today. In 2015, Apple announced plans to build...

Robotics foundation models have made remarkable progress. Today's best systems can follow natural language instructions to pick, place, sort, and manipulate a... Robotics foundation models have made remarkable progress. Today’s best systems can follow natural language instructions to pick, place, sort, and manipulate a wide variety of objects. But as these models grow more capable, evaluating them rigorously has become one of the field’s hardest unsolved problems. In this blog post, we introduce the key problems and our method for addressing them. Source

sqlite-utils 4.1 adds --code option to insert/upsert commands for inline Python data transformation.

ChatGPT is hiring a dedicated product manager to build experiences for families, caregivers, and older adults, according to a job posting.

Commentary noting an absence of major announcements following a week of model releases.

Meta told Dylan Byers, of Puck News, that it had nixed the feature after backlash from its user base.

Following significant backlash, Meta is turning off the feature it announced this week that let users generate AI images based on content from public Instagram accounts just by tagging them. The feature, as originally set up, meant that content from any public Instagram account could be used in AI creations without the account owner's permission. "Earlier this week, we announced that one way for people to generate images in Meta AI is by @-mentioning public Instagram accounts that they want to reference," Meta says in an update to a blog post about its new Muse Image AI model. "Our intent was...

Apple has sued OpenAI, alleging that former employees that now work for the AI company have stolen Apple's trade secrets "for the benefit of OpenAI." In its complaint, Apple alleges that it has uncovered "a pattern of theft of Apple's trade secrets by OpenAI employees who were formerly at Apple," and it names IO Products (Jony Ive's hardware startup that OpenAI bought in 2025), Tang Tan (OpenAI's chief hardware officer), and Chang Liu (who joined OpenAI from Apple in January) as defendants. An Apple spokesperson shared this statement with 9to5Mac: At Apple, our teams are constantly developing...

Apple alleges the misconduct was directed by OpenAi's senior leadership, including a long-time former employee.

Large language model (LLM) training workloads increasingly run into GPU memory limits before compute is fully used. Model weights, gradients, optimizer states,... Large language model (LLM) training workloads increasingly run into GPU memory limits before compute is fully used. Model weights, gradients, optimizer states, communication buffers, and intermediate activations all compete for GPU high-bandwidth memory (HBM). As model size, sequence length, and batch size grow, HBM capacity often becomes the primary scaling bottleneck. This post explains how… Source

Topological neural network (PHINN-EEG) for dream-state EEG classification using persistent homology; niche neuroscience application.

Visual pretraining framework for language models preserving document layouts and equations instead of text-only conversion.

Complex Social Behavior dataset benchmarks VLM error types across decade (2017-2025); tracks visual reasoning progression.

VEXAIoT: multi-agent LLM framework for autonomous IoT vulnerability discovery and exploitation testing.