Liberal MP says Australia risks sovereignty and strategic independence being ‘constrained by the AI superpowers reshaping the global order’

Employees are experiencing 'self-replacement' – silently withdrawing their judgment and feeling less valuable due to AI. This is more insidious than resistance. The article identifies three triggers (loss of control, trust, and legibility) and suggests fostering 'first wins' to reverse the trend.

Airis is a fully local, open-source AI ecosystem that runs with zero installation. It features an emotional core, screen analysis, mouse/keyboard automation, persistent memory, and mobile support, with comparisons to LM Studio and Open WebUI. The project emphasizes portability, no dependencies, and complete offline operation.

aurscan is a Go tool that scans AUR packages for malicious code before building, using Claude or local LLMs to analyze PKGBUILD files. It combines deterministic static rules with AI judgment to catch supply chain attacks like CHAOS RAT and Atomic Arch. Integrates with yay via a wrapper, supports multiple backends (Claude, API, local models), and fails closed for safety.

On June 13, 2026, a US government export control directive forced Anthropic to disable access to its two most powerful AI models for all foreign nationals, triggering global alarm about AI sovereignty and overreliance on American technology.

Tech companies are posting record profits while laying off tens of thousands, citing AI. Nearly 150,000 people have been affected this year, with layoffs occurring 44% faster than last year. Last month saw the highest monthly cuts in two years, and AI has been the top reason for layoffs across industries for three consecutive months.

The article examines the divide among AI critics: one side from institutional positions, nostalgic for past models and seeking to roll back; the other more radical, opposing AI and the structures that enabled it. The author argues the difference need not break unity but requires strategic communication.

This article explores the concept of token capital—the data, compute, and model assets that form the core of AI competitiveness—and how enterprises can build a continuous AI learning loop to leverage it. It highlights key strategies, benefits, and risks associated with owning token capital.

Document AI uses machine learning, NLP, and OCR to automatically extract, classify, and understand information from documents, turning them into structured data. Unlike traditional OCR, it understands context and meaning. Generative AI makes document AI more adaptable but still requires validation and human review. Governance is key for handling sensitive data.

Publia is a platform that ships what AI makes.

BrandLM provides one-time AI visibility audits across major AI platforms, showing brands how they appear in AI answers and where to improve.

A new report reveals that while 87% of digital workers use AI at work, only 13% say their organization's performance has significantly improved. Employees spend an average of 6.4 hours per week on 'botsitting'—checking, debugging, and cleaning up AI outputs. Moreover, 69% of AI users admit to 'botshitting'—shipping AI-generated work without thorough review. The report emphasizes that leading organizations are building the 'human infrastructure of AI' at individual, team, and organizational levels.

Z.ai launched GLM-5.2 on June 13, 2026, across all GLM Coding Plan tiers. The headline is a usable 1-million-token context window plus High and Max effort levels. It drops into Claude Code, Cline, and OpenClaw through an Anthropic-compatible endpoint. No benchmarks shipped at launch, and MIT open weights are promised next week.

Get paid to wait for Claude Code to finish.

Free AI Resume Builder that creates professional, ATS-friendly resumes without sign-up. Features AI-enhanced writing, import existing resume, one-click export to PDF or Word.

Applora is a free tool that uses AI to analyze negative reviews of Shopify apps, extracting real merchant pain points and delivering ranked opportunity briefs, helping developers discover market gaps backed by evidence.

AI is cutting customer service costs but accelerating organizational risk. Research shows AI chatbots hallucinate up to 82% on legal queries. When AI fails, brand Net Promoter Score can drop 70 points. This article explores three critical insights for deploying generative AI in customer service: trust thresholds as a deployment map, deterministic AI as a prerequisite for generative personalization, and escalation design as the measure of AI maturity.

AgentBridge is an open-source Python project that acts as a translation and governance mesh between different AI agent protocols. It supports MCP, A2A, ACP, OpenAI function-calling, Gemini, and AGNTCY, and provides identity, budgets, audit trails, and a policy engine. It's a working prototype with 6 protocols and 150+ tests passing.

A video presentation of Fugee, an AI assistant designed to help displaced people and asylum seekers navigate their new environments.

A state appellate panel upheld $6,000 in sanctions against a Southern California law firm’s attorneys for submitting a brief marred by generative AI mistakes in what a trial judge called “the worst example of misconduct by a lawyer that I think I’ve ever seen since I’ve been on the bench.”

A thought-provoking essay explores the question of whether using AI to complete school assignments constitutes cheating, delving into the educational system's dual goals of meritocratic sorting and formative development, and examining why students turn to AI in a high-stakes competitive environment.

MiMo Code is a coding agent with an explicit long-term memory architecture.

Researchers propose a reliability-aware diffusion planner that distills a vision-language model to generate scene-level reliability heatmaps, guiding UAVs to avoid unreliable regions (e.g., glass, mirrors) in 3D navigation, reducing obstacle violation rate from 40.3% to 9.6% and raising mean reliability from 0.588 to 0.925.

AnyGoal is a training-free multi-robot navigation architecture that uses a Vision-Language Model (VLM) for frontier-based exploration and coordinates agents via a shared 2D Gaussian Bayesian Value Map (BVM), achieving 52.4% subtask success rate on GOAT-Bench, a +27.5pp improvement over Modular GOAT.

ContactWorld benchmark spanning 12 contact-rich manipulation tasks reveals that spatially structured and temporally continuous representations, like point clouds, improve planning success rates from ~20% to 32.1%. Tactile sensing effectiveness depends on cross-modal compatibility; combining point clouds with tactile force fields achieves 36.1%. Tactile information becomes increasingly important for long-horizon planning.

A new study reveals the 'seed lottery' in single-GPU fine-tuning of vision-language-action models: out of 13 identical runs with different seeds, one silently degrades to 65.2% success rate from 91-94%. The culprit is output collapse, where the action predictor produces nearly identical outputs regardless of input. Weight-level regularizers fail to detect this, but output-level regularizers (VICReg, Dropout, or halved learning rate) eliminate all catastrophic seeds. The simplest fix is changing one number in the optimizer config.

This paper proposes an algorithm for efficient domain-adaptive policy learning using kernel representations. It models unknown disturbances with a differentiable random Fourier features kernel approximation. Offline training takes only 50 seconds on an RTX 4090 to optimize the policy via differentiable simulation. During deployment, the policy adapts in real-time by updating kernel parameters through online least-squares estimation. Experiments on Crazyflie quadrotors under various disturbances (wind, ground effects, payload shifts) validate the approach.

This survey examines the shift from isolated vehicle intelligence to multi-agent embodied systems in autonomous driving, focusing on Shared World Models (SWMs) as predictive cross-agent representations. Reviewing over 380 publications, it covers V2X communication, collaborative perception, inter-agent cognition, cooperative planning, end-to-end cooperative driving, and simulation engines. The study finds evaluation concentrated in simulation and offline protocols, with foundation-model-based coordination lacking real-time safety guarantees. Key research priorities include verifiable shared-state maintenance, robust intent and plan alignment, and safe coordinated action under communication and latency constraints.

World models in robot learning predict future states from visual observations and actions. FlowMo-WM infers object-centric motion and long-history context associated with hidden drift without direct flow supervision. It factorizes history into short-history latent and longer-history context, using zero-context residual transition to improve long-horizon prediction in simulated aquatic environments.

This paper presents $\mu_0$, a scalable world model that predicts 3D trajectories of interaction points (objects, tools, hands) rather than pixels or actions, enabling embodiment-agnostic robot learning. The TraceExtract system automatically extracts 3D supervision from videos. Experiments show $\mu_0$ outperforms baselines in trace prediction, and frozen $\mu_0$ can be paired with action experts for downstream tasks, achieving performance competitive with VLA models pretrained with action supervision.