Monitor
Anthropic barred the developer behind the OpenClaw jailbreak tool from using Claude, citing safety concerns over potential misuse.
Following U.S. warnings, Canadian regulators meet major financial institutions to address potential threats from the model.
AI's disruption of legacy software is driving a sharp sell-off in the sector, but investors see potential for AI-native efficiency plays to emerge as winners amid the chaos.
Experts predict 2026 as pivotal for AI world models achieving reliability, with key papers and prototypes advancing continual learning for more robust physical AI systems.
New engineering blog post outlines harnesses and infra for long-horizon AI agents, moving from prototype to enterprise deployment.
Medicare's new WISeR pilot program uses AI to scrutinize claims for overutilization, sparking concerns among providers and patients about potential delays in care approvals and increased denials.
DARPA is funding research to develop standards for AI agents to interoperate seamlessly, aiming to unlock multi-agent systems for complex tasks.
Anthropic is limiting Claude Mythos to vetted enterprises amid fears it could empower hackers, while firms race to patch flaws it uncovered.
The NYT details the closed-door session where officials highlighted Mythos' ability to find zero-days, drawing parallels to past cyber crises. Banks are now prioritizing AI-specific red-teaming.
A woman is suing OpenAI after ChatGPT allegedly reinforced her stalker's conspiracy theories despite her reports, raising questions about AI safety guardrails.
CoreWeave announced a major agreement to provide GPU cloud infrastructure for Anthropic's Claude family, becoming the platform for 9 of the top 10 AI model providers. The deal underscores exploding demand for AI compute amid capacity constraints. CoreWeave shares surged 11% on the news.
Meta's AI app reveals user activity to friends, leading to embarrassment and concerns over unintended social exposure.
Researchers at UC San Diego developed a novel chip that optimizes power conversion for GPUs, potentially slashing data center energy use amid AI boom.
OpenAI endorsed an Illinois bill limiting lawsuits against AI developers for catastrophic harms if they publish safety reports and avoid reckless actions. The bill targets frontier models over $100M compute, covering deaths, massive damage, or WMD development.
In his annual shareholder letter, Andy Jassy dismissed AI bubble fears, calling it a once-in-a-lifetime opportunity with AWS seeing explosive AI revenue growth. He highlighted strong customer commitments for AI infra capacity and faster adoption than past tech like electricity.
Cybersecurity stocks plunged after Anthropic's Mythos demo, with investors fearing AI will automate attacks faster than defenses evolve. CrowdStrike and peers lost billions in market cap amid broader sector selloff.
Industrial AI firm Radix launched RUSH 2026, a platform to deploy AI at scale in manufacturing and energy sectors, promising 20-30% efficiency gains via predictive maintenance and optimization.
AI-powered browser extensions are emerging as a stealthy, user-friendly vector for AI adoption, bypassing app stores and enabling seamless integration into daily workflows.
Snap's AR glasses unit Specs partners with Qualcomm on chips for upcoming AI-powered smart glasses, reviving consumer AR ambitions with on-device AI processing.
Anthropic's Claude AI identified a critical remote code execution vulnerability in Apache ActiveMQ that had evaded detection for over a decade, showcasing its superior cyber vuln hunting prowess.
Meta Platforms agreed to a massive expansion of its cloud deal with CoreWeave, totaling $21 billion, to fuel AI training amid surging datacenter needs.
Microsoft to invest more than $1 billion from 2026-2028 in Thailand for cloud and AI, including data centers and skills training. Aims to boost regional AI adoption. Part of broader Asia expansion.
Alibaba's ATH unit launched HappyHorse-1.0, topping Artificial Analysis leaderboards for text-to-video and image-to-video generation. The stealth model beat ByteDance's Seedance 2.0, with API coming April 30 amid competitors' setbacks like OpenAI's Sora shutdown.
RISC-V leader SiFive secured $400M in funding led by Nvidia and others to scale data center chip designs amid AI compute boom. Focus on custom AI accelerators challenging ARM dominance.