90 AI stories from this day, each in 60 words or fewer. Browse all days
The day in AI
Thursday, April 23
SpaceX secured a $60 billion option to acquire AI coding tool Cursor, while OpenAI plans a $1.5 billion private equity venture called DeployCo.
Google unveiled TPU v8 accelerators and a $750 million fund for agentic AI adoption during its Cloud Next 2026 event.
Anthropic signed a $25 billion infrastructure deal with AWS and released Claude Opus 4.7, its latest high-end model iteration.
Tencent and Alibaba are in investment talks with DeepSeek, valuing the AI startup at more than $20 billion.
The DHS demonstrated to lawmakers how to bypass safety guardrails in popular AI tools to plan terrorist attacks.
Lens Academy Launches Superintelligence 101 Course
AI Accelerates Thermoelectric Generator Design
OpenAI Launches ChatGPT for Clinicians
Anthropic Disputes Pentagon Supply Chain Risk Claims
Google Unveils TPU v8 AI Accelerators
AI Robot Outplays Humans At Table Tennis
SK Hynix Profits Jump Fivefold on AI Chips
Reese Witherspoon Defends AI Interest
DHS Demonstrates AI Jailbreaks For Terror Planning
Tencent And Alibaba Eye DeepSeek Investment
BluSky AI Joins Virtual Investor Conference
Native Communities Fight AI Data Center Expansion
Commvault Integrates Data Resilience Into Google Cloud
Spectral AI Showcases DeepView Burn Diagnostic System
Argument For Immediate AI Development Halt
Nightfood Holdings Scales Service Robotics
LLMs Suffer From Tool-Overuse Illusion
Researchers Find Ways To Reverse SynthID
Google Pledges $750 Million For AI Agents
Anthropic Restricts Mythos Over Cyber Security Risks
Vast Data Hits $30 Billion Valuation
Sony Robot Beats Pro Table Tennis Players
SpaceX May Acquire Cursor For $60 Billion
OpenAI Plans $1.5 Billion Private Equity Venture
Anthropic and Notion Update AI Interfaces
New Framework Addresses AI Proxy Failure
ZeroFolio Selects Algorithms Using Text Embeddings
SciBERT Tackles Class Imbalance in Science Scoring
New Framework Reduces LLM Hallucinations in AML Triage
ThermoQA Benchmarks LLM Thermodynamic Reasoning
Decoupling AI Consciousness From Moral Status
Romance Novelist Tests Claude's Cognitive Depth
Anthropic Signals New Claude Subscription Tiers
OpenAI Launches ChatGPT Images 2.0
MIT Tech Review Tracks Ten AI Trends
The Challenge Of Physical World Models
Deepfake Technology Enables Widespread Malicious Use
Google AI Research Debuts ReasoningBank
Community Misconduct Disputes Focus on Character, Not Facts
Photonic Computing for Deep Learning
Former OpenAI Researcher Launches Core Automation
Google Research Unveils AI Photo Re-composition
Gemma 4 VLA Debuts On Jetson Orin
OpenAI Offers Free ChatGPT For Clinicians
Tencent and Alibaba Eye DeepSeek Investment
Vista Equity Partners Partners With Google Cloud
California Mother Urges Chatbot Safety Regulations
Yelp Launches AI Recommendation Chatbot
Best Buy Names Jason Bonfig CEO
AI Agent Designs RISC-V CPU Core
Left-Leaning Groups Push AI Rules In Red States
UK Urges AI Defenses Against State Cyberattacks
Enterprise AI Requires Robust Data Fabric
New Framework Solves Variable Gapped LCS Problem
Simplifying Neural Computation in Superposition
Apple CEO Transition Focuses on AI Gap
Anthropic Claims Fair Use For Lyric Training
Secretary Duffy Limits AI in Air Traffic
Apple Presents New ML Research At ICLR
Combatting AI Ghostwriting in Academic Research
DeepMind Partners With Global Consultancies
Anthropic Releases Claude Opus 4.7
Anthropic Secures $25 Billion AWS Deal
Analyzing The Open-Closed Model Performance Gap
Huawei HiFloat4 Outperforms MXFP4 on Ascend Chips
GRASP Enables Long-Horizon World Model Planning
A Framework For Analyzing Open-Weight Models
Hacker News Bans AI-Generated Comments
Virtue Ethics Challenging AI Goal-Directedness
Common Pitfalls In Generative AI Development
Critique of Anthropic's Model Welfare Approach
The Third War of Priors Emerges
OpenAI Launches Autonomous Workspace Agents
PEER Sparse Experts Enable Int4 Training
Google Launches Three New AI Imaging Tools
Google Debuts 8th-Gen TPUs and Agent Platform
Unauthorized Users Access Anthropic Mythos Model
Meta Tracks Employee Keystrokes To Train Agents
SpaceX Secures $60 Billion Cursor Buy Option
European Chip Stocks Rally on AI Demand
OpenAI Uses WebSockets To Cut Agent Latency
GROVE Visualizes Language Model Output Distributions
ARES Framework Fixes Reward Model Failures
AI Scientists Fail Epistemic Reasoning Norms
Qutrit Neural Networks Outperform Qubits in Finance
New Framework For Recovering From Agent Errors
OpenAI Releases Open-Weight Privacy Filter Model
AI Robot Beats Elite Table Tennis Players
ChatGPT's Evolution Into An Everything App
LLMs Fuel Sophisticated Phishing Scams