Category intelligence

AI News Briefing — June 1, 2026

11 current items analyzed and ranked.

Executive synthesis

AI News Summary

Infrastructure & continual learning led the cycle. Trajectory, with UC Berkeley Sky Lab and Anyscale, open-sourced a concurrent multi-LoRA training stack reporting a 2.81× experiment-throughput gain for continual learning.

Research and adoption studies offered notable findings:

Startup Kaikaku.AI released Epicure, three models distinguishing recipe-based versus chemically-based ingredient pairings.

Agentic AI governance drew several practitioner-focused tools:

A tutorial implementing Microsoft's Agent Governance Toolkit for policy-routed, audited agent tool use.
SkillNet, a framework for discovering, evaluating, and organizing reusable agent skills.

Society and culture rounded out coverage: Anthropic banned AI tools in job interviews, Erin Brockovich launched a campaign against data center secrecy, and commentary addressed transhumanist ideology and AI psychosis debates.

Key Themes

AI Infrastructure and Data Centers · 3AI Research and Benchmarks · 3Agentic AI and Governance · 4AI Society and Culture · 3

Primary evidence

Top Ranked Signals

News The Decoder May 31 Old anchor

SoftBank plans 75 billion euro AI data center buildout in France

By Matthias Bastian

55 score

AI Analysis

SoftBank announced plans for up to 5 gigawatts of AI data center capacity in France, its largest European AI infrastructure investment at up to 75 billion euros, with 45 billion euros of facilities planned across three northern sites by 2031. The report notes SoftBank's pattern of large announcements that often fail to materialize.

SoftBank plans to build AI data centers with up to 5 gigawatts of capacity in France, the company's largest AI infrastructure investment in Europe, at up to 75 billion euros. By 2031, facilities worth 45 billion euros are set to go up at three sites in northern France. SoftBank's mega announcements keep stacking up worldwide, but many projects have yet to materialize. The article SoftBank plans 75 billion euro AI data center buildout in France appeared first on The Decoder.

AI infrastructureData centersInvestmentSoftBank

News The Decoder May 31 Old anchor

AI search agents often confirm what they already know instead of actually researching the web

By Jonathan Kemper

55 score

AI Analysis

Researchers at Harbin Institute of Technology introduced LiveBrowseComp, a benchmark using only events from the last 90 days, revealing that leading AI search agents including GPT-5.4 and Kimi K2.6 mostly confirm prior training knowledge rather than truly researching the web. When forced beyond memory, performance collapses and existing rankings shift dramatically.

Leading AI search agents like GPT-5.4 and Kimi K2.6 don't appear to do much actual research on established benchmarks. They mostly just use the web to confirm what they already learned during training. Researchers at the Harbin Institute of Technology found this using a new time-based benchmark called LiveBrowseComp, which only asks about events from the last 90 days. Once the models can't fall back on memory, performance falls apart and the existing rankings get reshuffled. The article

AI researchBenchmarksAI searchAgentic AI

News MarkTechPost May 31

Trajectory Releases a Concurrent Multi-LoRA Training Stack for Continual Learning, Reporting a 2.81× Experiment-Throughput Gain

By Michal Sutter

48 score

AI Analysis

Trajectory, working with UC Berkeley Sky Lab and Anyscale, released an open-source concurrent multi-LoRA training stack for continual learning, reporting a 2.81x experiment-throughput gain over single-tenant RL. The approach aims to replace discontinuous model release cycles with continuous learning, with all code in the NovaSky-AI/SkyRL repository.

Trajectory’s concurrent multi-LoRA stack reports a 2.81× experiment-throughput gain over single-tenant RL, with all code in the NovaSky-AI/SkyRL GitHub repository. Most language models improve in discontinuous jumps. A team collects data, trains, and ships a new version. This takes months and produces remarkable or catastrophic behavior for users. Trajectory wants to replace that cycle with continual learning. The Trajectory team published a field report describing how. It built a co

Open sourceContinual learningTraining infrastructureLoRA

News The Decoder May 31

Anthropic study finds men use AI coding agents more than twice as often as women in social science research

By Matthias Bastian

40 score

AI Analysis

An Anthropic study found that researchers with typically male names use AI coding agents more than twice as often as those with female names, even controlling for discipline and seniority. The gender gap is far wider for coding agents than for general AI use, varying sharply by field.

Researchers with typically male names use coding agents more than twice as often as those with typically female names, even within the same discipline and career level, according to an Anthropic study. Economists lead at 39 percent, while education researchers sit at just four percent. The gender gap for coding agents is far wider than for general AI use. The article Anthropic study finds men use AI coding agents more than twice as often as women in social science research appeared firs

AI researchAI adoptionCoding agentsEquity

News The Decoder May 31

Ask AI what goes with chicken and the answer depends on whether it learned from recipes or molecules

By Jonathan Kemper

38 score

AI Analysis

Startup Kaikaku.AI released Epicure, a set of three AI models that distinguish whether an ingredient suits a recipe or is chemically related, trained on millions of recipes and a flavor database. The chemistry-based variant surprisingly classifies taste and nutritional values better than recipe-trained models despite never seeing that data.

With "Epicure," London-based startup Kaikaku.AI presents three AI models that are the first to clearly separate whether an ingredient fits a recipe or is chemically related. Trained on 4.14 million recipes in seven languages and the FlavorDB flavor database, each variant returns different recommendations. The purely chemistry-based model even classifies taste and nutritional values better than the recipe-based alternatives, despite never seeing that information directly. The article Ask

AI researchApplied AIFood technology

News MarkTechPost May 31

An Implementation of the Microsoft Agent Governance Toolkit for Safe AI Agent Tool Use with Policies, Approvals, Audit Logs, and Risk Controls

By Sana Hassan

35 score

AI Analysis

A technical tutorial demonstrating how to build a governed AI agent workflow using Microsoft's Agent Governance Toolkit, routing every tool action through a policy layer with identity checks, risk tiers, approvals, audit logs, and a kill switch. It illustrates practical mechanisms for safe agentic tool use.

In this tutorial, we build a governed AI-agent workflow using Microsoft’s Agent Governance Toolkit as the reference point. We create a Colab-ready implementation where agents do not directly execute tools; instead, every action first passes through a governance layer that checks the agent’s identity, trust score, risk tier, requested tool, action type, sensitivity level, and policy rules. We define a YAML-based policy that controls destructive database operations, external email sending, shell e

Agentic AIAI safetyGovernanceTutorial

News The Decoder May 31

Anthropic bans AI tools during job interviews to see how candidates actually think

By Matthias Bastian

32 score

AI Analysis

Anthropic has banned AI tools during job interviews and uses up to five evaluation rounds testing skills, values, and ethical reasoning. The piece also notes very high salaries and the emergence of paid anonymous prep coaching by current AI company employees.

Anthropic bans AI during job interviews and runs candidates through up to five rounds testing skills, values, and ethical thinking. Salaries go up to $850,000, and some applicants pay $4,600 for prep coaching run anonymously by current AI company employees. The article Anthropic bans AI tools during job interviews to see how candidates actually think appeared first on The Decoder.

AI labsHiring practicesAnthropic

News AI News & Artificial Intelligence | TechCrunch May 31

Erin Brockovich takes aim at data center secrecy

By Anthony Ha

30 score

AI Analysis

Environmental activist Erin Brockovich is launching a campaign targeting the secrecy surrounding AI data center development. The effort highlights growing scrutiny over the environmental and community impact of AI infrastructure.

Environmental activist Erin Brockovich has a new mission.

Data centersAI policyEnvironment

News MarkTechPost May 31

Build Skill-Augmented AI Agents with SkillNet for Search, Evaluation, Graph Analysis, and Task Planning

By Sana Hassan

30 score

AI Analysis

A tutorial demonstrating SkillNet, a framework for discovering, installing, evaluating, and organizing reusable AI skills for agents, including semantic search, quality gating, and a skill-augmented task planner. It shows how complex goals can be decomposed into subtasks mapped to discovered skills.

In this tutorial, we implement a SkillNet use case as a practical framework for discovering, installing, inspecting, evaluating, and organizing reusable AI skills. We start by setting up a robust SkillNet client with SDK and REST fallback support, then compare keyword search with semantic search to understand how skills can be found for different task requirements. From there, we install curated skills from GitHub, inspect their metadata, apply a quality gate across key evaluation dimensions, an

Agentic AITutorialSoftware engineering

News AI (artificial intelligence) | The Guardian May 31

Our tech overlords are planning for conscious AI to conquer the cosmos. What could go wrong? | Eduardo Porter

By Eduardo Porter

22 score

AI Analysis

A Guardian opinion piece examining the transhumanist ideology spreading among wealthy tech leaders like Sam Altman and Elon Musk, who envision a future merging of humans and AI. It critiques the belief that humanity may design its own successor species and the cosmic ambitions surrounding conscious AI.

A new belief set is uniting some of the wealthiest men in the world around a ‘transhuman’ future – actual humanity be damnedSam Altman, the chief executive of OpenAI, took to the Internet a few years ago to propose that homo sapiens would be the first species “to design our own descendants”. In his best case scenario, the “merge” between humans and artificial intelligence occurs at some point over the next 50 years. The alternative, where we remain simply human and the machines follow their own

AI ideologyTech cultureTranshumanism

News AI News & Artificial Intelligence | TechCrunch May 31

Making sense of the debate over AI psychosis

By Anthony Ha

18 score

AI Analysis

A TechCrunch podcast episode debates the concept of AI psychosis and whether tech CEOs are particularly susceptible to it. The discussion explores psychological effects associated with heavy AI engagement.

On the latest episode of Equity, we debate whether tech CEOs are "uniquely prone to AI psychosis."

AI safetyMental healthTech culture