LLMs

Twitter

Facebook

Copy Link

Key Facts

Year2018

Key PlayersOpenAI, Google, Anthropic

Top News

Rebel Startups Challenge SK and LG in South Korea's 'AI Squid Game' for National AI Dominance

South Korea launched a state-backed AI competition dubbed 'AI Squid Game' where rebel startups Motif Technologies and Upstage face off against industrial giants SK Group and LG Group. The Ministry of Science and ICT oversees this multi-stage contest to select two best homegrown AI models, marking a shift in government strategy from backing conglomerates to supporting agile tech entrepreneurs in the global AI race.

Technology2 sources

Rebel Startups Challenge SK and LG in South Korea's 'AI Squid Game' for National AI Dominance

Technology2 sources

AI chatbots interpret 'likely' as 80% probability while humans assume 65%, new study reveals

A new study published in NPJ Complexity reveals that AI chatbots and humans interpret probability words differently, with language models assigning 'likely' to 80% while humans assume 65%. This probability misalignment poses risks in high-stakes fields like healthcare and government policy, where miscommunication about uncertainty could lead to flawed decisions.

Science2 sources

AI chatbots interpret 'likely' as 80% probability while humans assume 65%, new study reveals

Science2 sources

MatX AI chip startup raises $500 million to challenge Nvidia's dominance in LLM processing

MatX, an AI chip startup founded by former Google engineers, has secured $500 million in Series B funding led by Jane Street and Situational Awareness. The company aims to deliver processors 10 times better at training and running large language models than Nvidia's GPUs, with plans to begin shipping in 2027.

Startups4 sources

MatX AI chip startup raises $500 million to challenge Nvidia's dominance in LLM processing

Startups4 sources

CrowdStrike reports 89% surge in AI-driven cyberattacks as breakout times collapse to 29 minutes

CrowdStrike's 2026 Global Threat Report documents an 89% year-over-year increase in AI-enabled adversary operations, with average breakout time plummeting to just 29 minutes. Nation-state actors from Russia, China, and North Korea are deploying LLM-enabled malware and targeting enterprise AI systems directly, marking a fundamental shift in the cyber threat landscape.

Technology2 sources

CrowdStrike reports 89% surge in AI-driven cyberattacks as breakout times collapse to 29 minutes

Technology2 sources

ChatGPT Health fails to recognize half of medical emergencies in first independent safety test

OpenAI's ChatGPT Health missed over half of medical emergencies in a Nature Medicine study, directing patients to routine appointments instead of emergency rooms. With 40 million daily users seeking health guidance, the AI tool also showed alarming inconsistencies in suicide-crisis safeguards, triggering alerts for low-risk cases while failing to respond when users described specific self-harm plans.

Health9 sources

ChatGPT Health fails to recognize half of medical emergencies in first independent safety test

Health9 sources

Oura launches proprietary AI model for women's health in Oura Advisor chatbot

Oura introduces its first custom AI model designed specifically for women's health, enabling its Oura Advisor chatbot to provide personalized guidance on reproductive health. The model combines clinician-reviewed research with biometric data from the Oura Ring to address questions spanning menstrual cycles, fertility, pregnancy, and menopause. Currently rolling out in Oura Labs for opt-in testing, the feature marks a shift toward more specialized health AI.

Technology6 sources

Oura launches proprietary AI model for women's health in Oura Advisor chatbot

Technology6 sources

Firefox 148 rolls out AI kill switch as Mozilla balances user choice with modern AI browser vision

Mozilla released Firefox 148 with a comprehensive AI kill switch that lets users disable all AI enhancements with one click. The feature blocks ChatGPT integrations, sidebar chatbots, AI link previews, and smart tab suggestions. This move addresses user backlash while Mozilla pursues its AI-focused strategy under new leadership.

Technology3 sources

Firefox 148 rolls out AI kill switch as Mozilla balances user choice with modern AI browser vision

Technology3 sources

Russian hacker uses AI tools to breach 600+ FortiGate firewalls across 55 countries in 5 weeks

A Russian-speaking cybercriminal armed with commercial generative AI tools compromised more than 600 FortiGate firewalls across 55 countries between January and February 2026. AWS security researchers discovered the campaign relied on AI-generated scripts and automated attacks targeting weak credentials rather than sophisticated exploits, demonstrating how AI is lowering the barrier for cybercrime.

Technology7 sources

Russian hacker uses AI tools to breach 600+ FortiGate firewalls across 55 countries in 5 weeks

Technology7 sources

Tech Mahindra partners with University College London to advance Generative AI and quantum research

Tech Mahindra and University College London announced a collaboration to advance joint research in emerging technologies including Generative AI and quantum computing. The partnership will develop domain-specific Small Language Models, world models, and quantum machine learning solutions across healthcare, telecommunications, and manufacturing sectors.

Technology2 sources

Tech Mahindra partners with University College London to advance Generative AI and quantum research

Technology2 sources

ChatGPT cracks decades-old gluon amplitude puzzle, marking AI's first major theoretical physics win

ChatGPT solved a complex gluon amplitude problem that had stumped theoretical physicists for decades, proving that certain particle interactions previously thought impossible can actually occur. Working alongside Andrew Strominger and a team of researchers, the AI simplified complex mathematical expressions and generated a proof in just 12 hours—what might have taken humans indefinitely to solve.

Science2 sources

ChatGPT cracks decades-old gluon amplitude puzzle, marking AI's first major theoretical physics win

Science2 sources

Microsoft Pulls AI Training Guide After Backlash Over Pirated Harry Potter Books

Microsoft removed a developer blog post that instructed users on how to train large language models using pirated Harry Potter books. The post, written by senior product manager Pooja Kamath in November 2024, linked to a Kaggle dataset containing all seven novels incorrectly marked as public domain. The removal highlights growing ethical and legal challenges surrounding AI model training and the use of copyrighted material without permission.

Policy2 sources

Microsoft Pulls AI Training Guide After Backlash Over Pirated Harry Potter Books

Policy2 sources

AI Security Crisis: 61% of Companies See AI as Top Data Threat as Agents Gain Insider Access

A new global report reveals that 61% of organizations now identify AI as their primary data security risk, as AI agents gain broad access to enterprise systems with fewer controls than human workers. The research exposes a troubling gap: companies are granting AI tools insider-level privileges while 47% of sensitive cloud data remains unencrypted and nearly two-thirds have lost track of their data entirely.

Technology8 sources

AI Security Crisis: 61% of Companies See AI as Top Data Threat as Agents Gain Insider Access

Technology8 sources

Google Gemini 3.1 Pro doubles reasoning score, beats rivals in key AI benchmarks

Google released Gemini 3.1 Pro with dramatically improved problem-solving and reasoning capabilities, more than doubling its predecessor's performance on abstract reasoning tests. The newest and more powerful large language model scored 77.1% on ARC-AGI-2 and achieved a record 44.4% on Humanity's Last Exam, outperforming OpenAI and Anthropic in most benchmarks as accelerating competition among major tech companies intensifies.

Technology27 sources

Google Gemini 3.1 Pro doubles reasoning score, beats rivals in key AI benchmarks

Technology27 sources

India unveils homegrown AI models at major summit, but experts say DeepSeek moment remains distant

Indian AI startups showcased locally-built models at New Delhi's AI Impact Summit, with Prime Minister Modi celebrating 'Made in India' innovation. Sarvam AI released models optimized for 22 Indian languages, while Gnani.ai launched Vachana speech models trained on over a million hours of audio. But analysts suggest India's realistic path lies in becoming the world's largest AI adoption market rather than a frontier innovation hub.

Technology2 sources

India unveils homegrown AI models at major summit, but experts say DeepSeek moment remains distant

Technology2 sources

MIT Researchers Expose Hidden Biases and Personalities in Large Language Models

Researchers from MIT and UC San Diego have developed a method to identify and manipulate over 500 hidden concepts within large language models, including biases, personalities, and moods. The technique uses a Recursive Feature Machine algorithm to steer AI responses, improving performance while also exposing potential vulnerabilities like jailbreaking and hallucinations.

Science3 sources

MIT Researchers Expose Hidden Biases and Personalities in Large Language Models

Science3 sources

Prev 1…101112…23 Next