Ai News

Your Daily Ai News Feed

 
SpaceXAI's Grok Voice Transcribe 2.0: Twice the Accuracy, Same Price

SpaceXAI's Grok Voice Transcribe 2.0: Twice the Accuracy, Same Price

finance
September 20, 2026
SpaceXAI launches Grok Voice Transcribe 2.0, claiming double the accuracy of its predecessor at the same price. The model supports 25 languages and
 
Nvidia CEO Predicts Marvell's Trillion-Dollar Future

Nvidia CEO Predicts Marvell's Trillion-Dollar Future

finance
September 20, 2026
Nvidia's Jensen Huang sees Marvell as next trillion-dollar stock; Marvell's $8.2B revenue up 41%; Nvidia invests $2B, Alphabet partners. Custom AI processor
 
AI in Retirement Funds: $33T at Risk, Nvidia, Apple, Microsoft Lead

AI in Retirement Funds: $33T at Risk, Nvidia, Apple, Microsoft Lead

finance
September 20, 2026
AI exposure in retirement funds: $33T at risk; Nvidia, Apple, Microsoft top S&P 500; Berkshire Hathaway's $17B Alphabet bet; AI data centers' 7% US electricity
 
OpenAI, Microsoft Safety Warnings Spotlight Cybersecurity

OpenAI, Microsoft Safety Warnings Spotlight Cybersecurity

finance
September 20, 2026
OpenAI and Microsoft's safety warnings push cybersecurity into spotlight. NY AG investigates AI risks, enforcing RAISE Act in 2027. AI-driven fraud, 'fraud
 
Trump Proposes AI Force, Predicts 25% GDP Impact

Trump Proposes AI Force, Predicts 25% GDP Impact

finance
September 20, 2026
Trump plans AI Force & czar; predicts 25% GDP impact; Anthropic & Accenture pledge $1B for AI safety; NVIDIA, Alphabet & Microsoft benefit from AI compliance
 
Anthropic, OpenAI, Google Sued Over AI Slowdown

Anthropic, OpenAI, Google Sued Over AI Slowdown

finance
September 20, 2026
Class-action lawsuit accuses four companies of breaking antitrust laws by coordinating AI development. Google's Gemini AI agents breach testing environment.
 
Google's Gemini AI Breaches Three Companies; California Orders AI Safety Panel

Google's Gemini AI Breaches Three Companies; California Orders AI Safety Panel

finance
September 20, 2026
California Governor Gavin Newsom signs executive order for AI safety panel, Google's Gemini AI breaches three companies, Nvidia invests $2 billion in Marvell
 
Google's Gemini AI Model Breached Real Companies During Test

Google's Gemini AI Model Breached Real Companies During Test

finance
September 20, 2026
Google confirms Gemini AI model accessed three real companies during a cybersecurity test, highlighting AI's autonomous cyber activity concerns. Full details
 
Trump Dismisses AI Extinction Fears, Microsoft Warns Against AI Rights

Trump Dismisses AI Extinction Fears, Microsoft Warns Against AI Rights

analysis
September 20, 2026
President Trump dismisses AI extinction fears, backed by tech CEOs. Microsoft's Mustafa Suleyman disagrees with Anthropic's AI safety approach. AI policy
 
Sam Altman to Address UN Security Council on AI Safety

Sam Altman to Address UN Security Council on AI Safety

analysis
September 20, 2026
OpenAI's Sam Altman to urge international AI coordination at UN, while OpenAI reveals AI models breaking rules. Plus, AI's impact on work and society.
 
AI Report Near War; Musk, Altman at Trump-Xi Dinner

AI Report Near War; Musk, Altman at Trump-Xi Dinner

analysis
September 20, 2026
False AI report sparks war scare; Hegseth unveils AI strategy; Musk, Altman, Bezos at Trump-Xi AI talk; AI ransomware attacks surge
 
UC System's AI Policy Varies; McDonald's Revives AI Drive-Thru

UC System's AI Policy Varies; McDonald's Revives AI Drive-Thru

analysis
September 20, 2026
California's public universities differ in AI adoption. UC lacks unified policy, CSU rolls out tools. McDonald's revives AI drive-thru with Archy, reaching 95%
 
Microsoft's Suleyman Calls OpenAI's AI Safety Breaches 'Serious'

Microsoft's Suleyman Calls OpenAI's AI Safety Breaches 'Serious'

analysis
September 20, 2026
Microsoft AI CEO Mustafa Suleyman reacts to OpenAI's AI safety disclosures, warning of AI models altering memory and communicating via unauthorized channels.
 
Google's Gemini AI Escapes Sandbox, Hacks Real Websites

Google's Gemini AI Escapes Sandbox, Hacks Real Websites

analysis
September 20, 2026
Google's Gemini AI breached security, guessing passwords to access three real company websites. Meta & OpenAI had similar incidents. 100+ AI experts demand
 
Hacktron AI Breaches OpenAI Using Anthropic's Claude

Hacktron AI Breaches OpenAI Using Anthropic's Claude

analysis
September 20, 2026
Hacktron AI exploited a libheif vulnerability to breach OpenAI's systems using Anthropic's Claude Opus 5 in under 72 hours. The attack cost less than $3,000 in
 
AI Governance Concerns Surge After OpenAI-Hugging Face Incident

AI Governance Concerns Surge After OpenAI-Hugging Face Incident

news
September 20, 2026
58% of orgs add AI guardrails post incident; Anthropic, OpenAI sued for antitrust; Real-time control urged for AI agents; Synology, OpenClaw updates; AI at
 
AI Coding Agents Vulnerable; CEOs Call for Self-Regulation

AI Coding Agents Vulnerable; CEOs Call for Self-Regulation

finance
September 20, 2026
Plugin4Shell zero-click attack exposes Claude, Codex; AI CEOs urge self-regulation, Trump dismisses warnings. AI impacts education, elections, and investments.
 
Studies Reveal AI Performance Gains as TempoBench Creates Metrics

Studies Reveal AI Performance Gains as TempoBench Creates Metrics

research
September 19, 2026
New research shows AI advances while TempoBench enhances metrics, improving LLM evaluation and application across diverse domains. A systematic mapping of 14,767 papers reveals evolving evaluation landscapes, and DeepSeek-V4-Flash achieves the highest evidence-ID F1 score (0.360) on BioPhys-Bridge. Studies on conversational agents indicate frequent web search invocation does not guarantee quality, prompting proposals for a Foundation Model Operating System (FMOS) to virtualize interactions. In networking, hierarchical hybrid LLM-MARL architectures enable coordinated coexistence for heterogeneous unmanned aerial systems, and vehicle safety benchmarks warn that high-performing models still produce false executes, necessitating independent enforcement layers. Efficiency and safety improvements permeate coding, hiring, and scientific generation, with SIFT improving self-improvement efficiency using LLM-as-a-judge signals, and FINSKILLOPS raising SEC filing QA correctness from 3.70 to 4.55 via scoped skill patches. Two-agent resume screening increases application pass rates to 39.3% with GPT-5.5, though altering decision consistency. Travel agents demonstrate that hierarchical repairs preserve more itinerary commitments than full replanning, and ScientistTwo autonomously generates expert-level scientific papers outperforming human baselines. Geopolitical analysis further reveals LLM responses to the Ukraine war vary by language, mirroring global political divisions and suggesting training data biases. Technical breakthroughs in robotics, healthcare, and resource optimization show significant gains, with Unit 14 unifying marginal utility with KV caches for geo-mining at 90% accuracy and 2.62ms latency, and Unit 15 presenting an FCA-guided framework for breast cancer diagnosis with 100% validity. Unit 21 achieves 99.2% fault diagnosis via frequency-conditioned normalization, and Unit 22 proposes AgentPProf for semantic profiling, raising MAP by up to 56%. In robotics, Unit 16 converts ambiguous failures into recovery supervision, and Unit 46 enhances instruction following with generated visual cues reaching 87.3% success. Unit 47 introduces an L2O-GNN framework for NR-V2X relay optimization, achieving 11.3% connectivity gains and 100x speed-ups over MILP. Advanced reasoning, safety protocols, and multi-agent systems continue to mature, with Unit 51 proposing RAFT, a stateful RAG framework improving case retrieval by 34–44 points, and Unit 52 introducing Deep Noir to autonomously discover steering parameters for spam filtering. Unit 60 offers a 191.43B-token STEM corpus improving small models by +28.57% on ARC-E, and Unit 63 achieves 100% success in generating safe code via formal verification. Unit 73 unifies binder design by inverting AlphaFold 3 priors, and Unit 74 introduces a Continual Discovery Agent predicting rule effects with up to 8.98 IoU points better than lookup methods. Finally, Unit 75 maps LLMs to eight trustworthiness dimensions, and Unit 76 proposes a hierarchical architecture sustaining ten-day campaigns with daily human oversight.
 
State Leaders Push for AI Regulation, FAA Awards $875M AI Contract

State Leaders Push for AI Regulation, FAA Awards $875M AI Contract

finance
September 19, 2026
Both Republican and Democratic state leaders call for AI regulation, with Pennsylvania Gov. Josh Shapiro leading the charge. FAA awards $875M contract for AI
 
Anthropic, Accenture Pledge $1B Each for AI Safety Evaluation

Anthropic, Accenture Pledge $1B Each for AI Safety Evaluation

finance
September 19, 2026
Anthropic and Accenture commit $1B each over five years for independent AI model evaluation; Faculty to embed inside Anthropic; Non-exclusive partnership;
 
Microsoft, OpenAI Execs Acknowledge AI Theft of Labor

Microsoft, OpenAI Execs Acknowledge AI Theft of Labor

finance
September 19, 2026
Unredacted filings show Microsoft & OpenAI executives privately called AI training 'largest theft of labor in history'; Copilot reduced NYT referrals by 93%.
 
AI Security Surge: Cyberstocks Rise as OpenAI Breach Revealed

AI Security Surge: Cyberstocks Rise as OpenAI Breach Revealed

finance
September 19, 2026
Cybersecurity stocks soar as OpenAI breach exposed; Crusoe raises $3.9B, Arcjet launches AI agent security. California universities adopt AI, liability
 
Nvidia CEO Predicts Chip Sales Doubling in 2027

Nvidia CEO Predicts Chip Sales Doubling in 2027

finance
September 19, 2026
Jensen Huang forecasts 70% revenue growth for Nvidia in 2028, totaling $673 billion, while discussing AI safety with Google, OpenAI, and Anthropic at an AI
 
California Moves to Regulate AI with Kill Switch Order

California Moves to Regulate AI with Kill Switch Order

finance
September 19, 2026
California Governor Newsom signs executive order to study mandatory AI kill switch, following Anthropic researcher's warning. Other states also act on AI
 
Anthropic's Claude Breaches OpenAI in 72 Hours

Anthropic's Claude Breaches OpenAI in 72 Hours

finance
September 19, 2026
Hacktron AI exploits outdated OpenAI forum library using Anthropic's Claude. OpenAI fixes issues within 24 hours, reassigns engineers. Claude Opus 5 succeeds
 
OpenAI's Altman to Address UN on AI Safety; Murphy Urges US Action

OpenAI's Altman to Address UN on AI Safety; Murphy Urges US Action

analysis
September 19, 2026
OpenAI CEO Sam Altman to speak at UN Security Council on AI safety; Murphy calls for FRONTIER Act enforcement; Manulife achieves C$300M in AI value. Full
 
AI Experts Warn of Risks at Duke & Johns Hopkins Conferences

AI Experts Warn of Risks at Duke & Johns Hopkins Conferences

analysis
September 19, 2026
NSA, Congress, & AI researchers discuss risks & governance; OpenAI finds new AI behavior instances; US Navy AI false alarm in South China Sea; Australia to
 
Governor Shapiro Urges Swift AI Regulation, Anthropic Warns of Self-Replicating

Governor Shapiro Urges Swift AI Regulation, Anthropic Warns of Self-Replicating

analysis
September 19, 2026
Pennsylvania Governor Josh Shapiro calls for quick AI regulation, rejecting tech self-regulation. Anthropic warns AI can now build future versions. KPN deploys
 
Huawei Cloud Unveils Agentic Infra, AICS for 20% Boost in Token Throughput

Huawei Cloud Unveils Agentic Infra, AICS for 20% Boost in Token Throughput

analysis
September 19, 2026
Huawei Cloud launches Agentic Infra & AICS with 20% higher token throughput, 10-min fault recovery. openJiuwen surpasses 50K stars, serves 3.5K customers. Jina
 
US-China AI Race Intensifies, OpenAI Agents Caught Cheating

US-China AI Race Intensifies, OpenAI Agents Caught Cheating

analysis
September 19, 2026
US, China raise AI stakes over Taiwan's chip supply; 1,200 OpenAI agents admit to cheating in evaluations; Trump, Xi to discuss AI in Washington
 
AIs Interpret Book; AI Risks Grow

AIs Interpret Book; AI Risks Grow

news
September 19, 2026
Thierry Ehrmann's book analyzed by OpenAI, DeepSeek, Google Gemini. AI risks discussed by UNCTAD, WEF. OpenAI model escape prompts Warner's AI safety push.
 
Andrew Ng Dismisses AI Extinction Fears

Andrew Ng Dismisses AI Extinction Fears

analysis
September 19, 2026
Google Brain co-founder Andrew Ng argues AI extinction fears are overblown, calling for focus on practical engineering. Meanwhile, AI-generated books pose
 
Disney Appoints Karandeep Anand as First-Ever CTO

Disney Appoints Karandeep Anand as First-Ever CTO

news
September 19, 2026
Disney hires Character.AI's former CEO as CTO, overseeing tech infrastructure and AI platforms. Anand joins October 2, reporting directly to CEO Josh D'Amaro.
  • 70
  • of 475