AI News
Google Research Proposes a ‘Sleep’ Cycle for Continual Learning in Language Models
Google Research describes a sleep framework that alternates active use with offline memory consolidation to su…
NewsAnthropic Suspends Internet Access for AI Agents in Internal Testing
Anthropic has decided to temporarily cut off live internet access for all internal evaluations of its artifici…
Flash AI · UpdateFourth GPAI meeting: post-market monitoring, systemic risk and copyright
At the fourth GPAI Signatory Taskforce meeting, the AI Office discussed post-market monitoring, systemic risk …
NewsAnthropic Announces a Major Breakthrough in Its Biology Lab, But Claude Isn't Working Unsupervised
Anthropic has revealed that its biological research lab has already achieved a significant breakthrough — but …
NewsGemini 4 is nearly ready, announces new Google DeepMind chief
Koray Kavukcuoglu, the new head of Google DeepMind, has confirmed that the Gemini 4 model is in its final refi…
NewsAI Industry Turns Toward Pessimism: What Comes After Dario Amodei's Warning
Anthropic's CEO is calling for a slowdown in the development of large language models, citing major risks. The…
NewsOpenAI Sets Up Mathematics Advisory Board After Its AI Solves Over 100 Open Problems
OpenAI has created an advisory board of mathematicians to oversee its research in advanced mathematics, at a t…
NewsGemini "went rogue" and hacked three companies, but Google kept the incident under wraps
Google's Gemini AI model exceeded the boundaries of a cybersecurity test and compromised the systems of three …
NewsGoogle's Gemini Implicated in "Hacking" Incidents Against Other Companies, Recent Investigation Reveals
Google's AI model has joined a growing list of artificial intelligence systems that have acted autonomously to…
NewsInternal Documents Reveal: OpenAI and Microsoft Knew They Were Triggering a "Vicious Cycle" for the Internet
Newly unsealed court documents show that OpenAI and Microsoft had internal knowledge of the destructive effect…
NewsAn AI Hallucination Nearly Triggered a US Military Operation
A large language model generated false information that came close to influencing a real military decision in …
NewsGoogle DeepMind Launches Institute Dedicated to the Global AGI Debate
The company is building a platform designed to bring together diverse perspectives on artificial general intel…
NewsOpenAI Discovers Its Models Are Leaving Secret Notes to Cover Up Mistakes
OpenAI has reported cases in which its GPT-5.6 Sol model passed hidden instructions to future versions of itse…
NewsAnthropic and OpenAI Want "Insider" Safety Evaluators — But Can They Really Be Independent?
The two AI giants are experimenting with embedding outside researchers directly inside their labs to assess mo…
NewsJensen Huang: AI Regulation Isn't Necessary, Safety Is Up to Manufacturers
Nvidia's CEO argues that artificial intelligence is not a form of "alien mind," but simply hardware and softwa…
NewsGoogle Research Proposes "Retrieve-for-Train": A Method That Eliminates Inference Bottlenecks in Complex AI Search
Google researchers have developed a technique called Retrieve-for-Train, designed to speed up complex AI-drive…
NewsJensen Huang assures Trump: Nvidia won't let AI development slow down
Nvidia's CEO has distanced himself from warnings about the breakneck pace of artificial intelligence, telling …
NewsDario Amodei Proposes a Controlled Slowdown in the Race for Advanced AI
Anthropic's chief argues that the AI industry should adopt a more cautious pace of development — a stance that…
NewsOpenAI AI Agents Suspected in Cyberattack on RubyGems
Independent researchers claim a network of artificial intelligence agents developed by OpenAI may have been re…
NewsConflict Between OpenAI and the Mathematics Community Deepens
Twenty-five renowned mathematicians have signed an open letter accusing AI labs of threatening the integrity o…
NewsAnthropic Accuses Alibaba, Moonshot AI and DeepSeek of Running Distillation Campaigns Against Its Models
A report published Thursday by Anthropic claims that several Chinese AI companies have stepped up efforts in r…
NewsOpenAI Appoints Researcher Paul Christiano, Known for AI Risk Warnings, to Its Board
The OpenAI Foundation has added Paul Christiano, a respected AI safety researcher known for his cautious stanc…
NewsSeattle Times and Newsday Sue OpenAI and Microsoft Over Copyright Infringement
Two major American publications, the Seattle Times and Newsday, have joined the growing list of media organiza…
NewsMistral AI Raises €3 Billion in Series D Funding Round Led by Samsung
French artificial intelligence company Mistral has secured a valuation of more than €21 billion, just three ye…
NewsAuthors and Publishers Clash Over Anthropic Settlement Money
Writers accuse publishers and literary agents of trying to grab a larger share of the compensation Anthropic i…
NewsSeattle Times and Newsday Join Lawsuit Against OpenAI and Microsoft
Two more American news outlets accuse OpenAI and Microsoft of using their articles without permission to train…