Wiredframe Radar logo

Category intelligence

AI News Briefing — May 25, 2026

8 current items analyzed and ranked.

Executive synthesis

AI News Summary

Research and model releases dominate the frontier signal this cycle:

  • Microsoft Research open-sourced Webwright, a terminal-native web agent that writes Playwright code instead of clicking actions, scoring 60.1% on Odysseys versus 33.5% for base GPT-5.4.
  • NVIDIA released Gated DeltaNet-2, a linear attention layer decoupling erase/write in the delta rule, beating Mamba-2, Mamba-3, and KDA at 1.3B parameters on 100B tokens.
  • StepFun launched StepAudio 2.5 Realtime, an end-to-end voice LLM with roleplay-specific RLHF, paralinguistic comprehension, and million-scale persona data augmentation.

Policy, industry, and cultural threads highlight tensions around AI's footprint and hype:

Key Themes

Model and Architecture Releases · 3AI Infrastructure and Sustainability · 1AI Hype and Society · 3LLM Tooling · 1

Primary evidence

Top Ranked Signals

News MarkTechPost May 24

Microsoft Research Releases Webwright: A Terminal-Native Web Agent Framework That Scores 60.1% on Odysseys, Up from Base GPT-5.4’s 33.5%

By Asif Razzaq

85 score
AI Analysis

Microsoft Research released Webwright, an open-source terminal-native web agent framework that lets agents write Playwright code instead of taking single browser actions. It scores 60.1% on Odysseys benchmark versus base GPT-5.4's 33.5%.

Most web agents today drive a browser one action at a time. The model receives the current page state — as a screenshot or DOM text — and predicts the next click, keypress, or scroll. This action-at-a-time design made sense when language models had limited reasoning ability. As models have become more capable at writing and debugging code, that rigid loop has become a constraint rather than a structure that helps. Microsoft Research’s AI Frontiers lab built a different approach. Their n
AI agentsOpen sourceMicrosoft Research
News MarkTechPost May 24

NVIDIA AI Releases Gated DeltaNet-2: A Linear Attention Layer That Decouples Erase and Write in the Delta Rule

By Asif Razzaq

82 score
AI Analysis

NVIDIA released Gated DeltaNet-2, a linear attention layer that decouples erase and write operations in the delta rule via channel-wise gates. Trained at 1.3B parameters on 100B FineWeb-Edu tokens, it outperforms Mamba-2, Mamba-3, Gated DeltaNet, and KDA.

Linear attention replaces the unbounded KV cache of softmax attention with a fixed-size recurrent state. This cuts sequence mixing to linear time and decoding to constant memory. The hard part is not what to forget. It is how to edit a compressed memory without scrambling existing associations. NVIDIA has released Gated DeltaNet-2, a linear attention layer that targets that bottleneck. The model decouples the active memory edit into two channel-wise gates. It is trained at 1.3B parameters on
Model architectureLinear attentionNVIDIA research
News MarkTechPost May 24

StepFun Releases StepAudio 2.5 Realtime: An End-to-End Voice Model with Roleplay-Specific RLHF and Paralinguistic Comprehension

By Michal Sutter

75 score
AI Analysis

StepFun released StepAudio 2.5 Realtime, an end-to-end real-time speech LLM with customizable persona capabilities, roleplay-specific RLHF, and paralinguistic comprehension. It supports Chinese and English via a WebSocket API and uses million-scale persona data augmentation.

StepFun, the Shanghai-based AI lab, released StepAudio 2.5 Realtime. It is an end-to-end real-time speech large language model with fully customizable persona capabilities. StepAudio 2.5 Realtime is a voice model that operates in real time. Unlike pipeline-based systems that separate speech recognition, reasoning, and synthesis into sequential steps, this is an end-to-end model. Audio goes in and audio comes out through a single unified system. The model supports Chinese and English. It c
Voice AIModel releaseMultimodal
News AI (artificial intelligence) | The Guardian May 24

Scotland’s ‘green datacentres’ policy ignores emissions impact of AI, analysis shows

By Aisha Down

55 score
AI Analysis

A Scottish charity analysis warns that Scotland's 'green datacentres' policy, defined in 2022 before ChatGPT, ignores the massive carbon emissions impact of AI workloads. The policy underpins UK efforts to attract AI investment but may mask significant environmental costs.

Definition of green facilities made in 2022, before release of ChatGPT, says Action to Protect Rural ScotlandA Scottish government policy designed to encourage datacentres to build in Scotland could lead to a massive volume of carbon emissions being ignored, according to an analysis by a Scottish charity.“Green datacentres” are at the heart of Scotland’s ambitions to develop economically. Enshrined in national policy, they are part of a larger, UK-wide effort to attract big AI investment to Scot
AI infrastructureSustainabilityPolicy
News AI (artificial intelligence) | The Guardian May 24

‘We’re expanding the cinematic toolbox’: AI fault lines on show at Cannes

By Nadia Khomami Arts and culture correspondent

50 score
AI Analysis

At Cannes, divisions over generative AI in filmmaking deepened, with Darren Aronofsky championing AI through his studio Primordial Soup while Guillermo del Toro voiced strong opposition. The festival highlighted AI as Hollywood's most divisive issue.

Darren Aronofsky among proponents of using technology, while Guillermo del Toro says he would ‘rather die’Under a white marquee on Cannes’ Croisette beach, with the Mediterranean glistening behind him and superyachts drifting across the horizon, the director Darren Aronofsky addressed an audience of executives and tech evangelists gathered for an “AI for Talent” summit.“There’s so much pushback against AI,” said Aronofsky, who has faced criticism over his embrace of generative AI projects though
Generative AICreative industriesFilm
News AI (artificial intelligence) | The Guardian May 24

‘AI washing’: firms are scrambling to rebrand themselves as tech-focused

By Aisha Down

45 score
AI Analysis

PR firms report that UK companies are aggressively rebranding ordinary automation as AI to capitalize on hype, a phenomenon dubbed 'AI washing.' Communications executives describe the pressure as 'yoga-level' stretching of the truth.

PR executives say UK companies are forcing them to present ordinary automation as artificial intelligence UK companies are performing “yoga-level” stretches to describe themselves as AI specialists in an attempt to capitalise on the buzz around the technology, public relations firms have said.Weary communications executives tasked with securing media coverage for brands have complained that bosses in low-tech industries or running businesses that use automation but not generative AI, are increas
AI hypeMarketingIndustry trends
News MarkTechPost May 24

Build a Complete Langfuse Observability and Evaluation Pipeline for Tracing, Prompt Management, Scoring, and Experiments

By Sana Hassan

40 score
AI Analysis

A tutorial on building a complete observability and evaluation pipeline using Langfuse, covering tracing, prompt management, scoring, datasets, and experiments for LLM applications.

In this tutorial, we implement the Langfuse (an open-source LLM engineering platform) pipeline for tracing, prompt management, scoring, datasets, and experiments. We build a complete workflow that works with either a real OpenAI key or a deterministic mock LLM, so we can understand every major Langfuse feature without depending on paid model access. We start by setting up credentials and connecting to Langfuse. We trace simple function calls, instrument a small RAG pipeline, manage prompts centr
LLM toolingObservabilityTutorial
News AI (artificial intelligence) | The Guardian May 24

I avoid AI tools because thinking is supposed to be hard. It’s what makes us human | Wendy Liu

By Wendy Liu

30 score
AI Analysis

An opinion piece arguing against AI tools, framing intellectual struggle as essential to human cognition and warning about big tech's privatization of intelligence.

As intelligence itself becomes privatised by big tech, allowing your intellectual faculties to wither in service of inane bots seems a dangerous moveLong before the age of multi-billion-dollar AI companies promising to disrupt the field of software development, I was learning to code the hard way.It was the mid-2000s, and I was a child with unmonitored access to the family computer. With the help of a basic text editor program, I learned how to make websites – first basic, then increasingly comp
AI ethicsSociety