All AI News

Access models from Anthropic and OpenAI are redefining AI cybersecurity deployment, shifting focus from raw model performance to controlled, defensive distribution. Organizations are now prioritizing secure, restricted access to prevent adversarial misuse.

Guangzhou Automobile Group’s Waymo-derived L4 algorithm, paired with full-system LiDAR, is now available at an unprecedented price point of 115,800 RMB—just 200 TOPS of compute power. This marks a seismic shift in autonomous driving accessibility.

Discover the top Claude shortcuts that can 10X your AI speed, from Fast Mode to project memory features. Move beyond basic prompts and unlock enterprise-grade efficiency.

China has secured its largest-ever investment in embodied AI with a $455 million round led by Hillhouse Capital and Sequoia Capital, signaling a strategic pivot toward full-stack AI brains. The funding aims to accelerate the development of autonomous, perception-driven robotic systems.

Gemini Robotics ER 1.6 introduces groundbreaking spatial reasoning and label identification, enabling robots like Boston Dynamics' Spot to navigate complex environments with human-like intuition. The update marks a leap in embodied AI, combining Google’s AI prowess with real-world robotics.

Tiled VAE and DiT support have been added to FaceDetailer and SeedVR2, dramatically reducing VRAM spikes in high-resolution Stable Diffusion workflows. This breakthrough enables smoother multi-model pipelines without frequent model reloading.

Finance leaders are raising alarms over the imminent deployment of Claude Opus 4.7 in UK banks, a powerful AI tool previously restricted to US firms. The model, developed by Anthropic, promises enhanced efficiency but poses unprecedented risks to financial security and regulatory compliance.

Salesforce has unveiled Headless 360, a new infrastructure layer allowing AI agents to interact with its entire ecosystem without requiring human login. This shift redefines CRM automation by enabling autonomous, agent-driven workflows across sales, service, and marketing.

Despite the rise of Mixture-of-Experts architectures, dense models continue to play a critical role in edge deployment and model efficiency. Recent research highlights their enduring value through distillation techniques and lightweight designs like TinyAya.

Claude Opus, Anthropic’s public AI model, successfully developed a Chrome exploit for a $2,283 bug bounty—revealing the power of accessible AI in cybersecurity. This comes as the company admits its restricted Mythos model remains far more potent.

The Qwen Team has open-sourced Qwen3.6-35B-A3B, a sparse Mixture-of-Experts vision-language model with only 3B active parameters and advanced agentic coding abilities. According to OfficeChai, it outperforms Google’s Gemma 4-31B on multiple benchmarks.

RoboChallenge has completed its second wave of ecosystem expansion, bringing together 18 of the world's most advanced embodied AI firms to create the largest competitive arena for embodied models. This unprecedented coalition signals a major leap in real-world AI deployment.

China has launched a high-level investigation into Meta’s $2 billion Manus deal, raising national security alarms. The probe examines potential data flows and foreign influence tied to the transaction.

SoftBank is set to launch the Natural AI Phone, a new smartphone touting predictive AI that anticipates user behavior, in a one-year exclusive domestic rollout. The device marks a major shift in consumer mobile technology in Japan.

The UK government has launched its first £500m sovereign AI fund, with Technology Secretary Liz Kendall urging public support despite growing fears over job displacement and AI-driven cyber threats.

AI has an awful image problem as growing public skepticism stems from tech giants failing to demonstrate tangible benefits. Without clear communication, innovation is perceived as threatening rather than transformative.

Crypto and AI PACs have raised over $250 million in campaign funding ahead of the US midterm elections, led by tech titans Marc Andreessen and Ben Horowitz, who contributed $25 million to a pro-AI super PAC in the first quarter alone.

Claude Mythos Preview, Anthropic’s latest AI model, can autonomously identify and exploit previously unknown vulnerabilities in financial systems, prompting urgent warnings from global regulators and cybersecurity experts.

Data centre delays are threatening to choke AI expansion across the US, with nearly 40% of projects facing hold-ups. Microsoft’s pivot to in-house models and takeover of OpenAI’s abandoned Texas site highlight growing strain on infrastructure.

Claude Opus 4.7 showcases breakthroughs in coding, data visualization, and reasoning—but its massive token usage raises efficiency concerns. According to Decrypt and Anthropic, this model pushes boundaries while demanding significant computational resources.

A new framework called LAMO enables lightweight multimodal language models to perform complex GUI automation through role-based orchestration, overcoming traditional scalability and cost barriers. This breakthrough bridges the gap between resource-constrained devices and advanced agent systems.

A new agentic AI framework delivers a safe, lightweight, and user-friendly workflow for autonomous scientific tasks, minimizing human intervention while maximizing reliability. Built on isolated execution and self-assessing loops, it represents a major leap toward practical AI-driven research.

Claude Opus 4.7 delivers breakthrough improvements in coding precision and adherence to user instructions, enabling developers to delegate complex tasks with unprecedented confidence. According to Anthropic, the model excels in long-running engineering workflows and self-verification.

Claude Opus 4.7 sets a new standard in AI performance, outperforming GPT-5.4 and Gemini 3.1 Pro in coding benchmarks and agentic workflows. Anthropic's latest update refines reasoning, tool use, and memory capabilities with unprecedented precision.