The Decoder

344 articles in the TruthFoundry index. Each links out to the original.

• Anthropic Claude Science produces first complete ultraviolet sky map · Anthropic's Claude Science AI created the first full ultraviolet map of the entire sky.
• OpenAI Exposes Russian and Iranian Influence Ops Using Fake Identities · OpenAI identified Russian and Iranian operations that planted fake stories in real news outlets using fabricated identities.
• Anthropic launches live dashboards and animated video features for Claude · Anthropic introduces Dashboards for live data visualization and Motion for AI-generated video in its Claude platform.
• Anthropic Bans Systematic Abuse of Claude Under New Usage Policy · Anthropic updated its usage policy to prohibit systematic abuse of the Claude AI model.
• Teen rescued from mountain after using AI for route planning · A 16-year-old climber was airlifted after relying on an AI chatbot to plan a dangerous mountain route.
• AI-Powered Tool Enables Breach of Multiple South Korean Banks · A suspected attacker used an AI-driven tool to steal data from multiple South Korean financial institutions.
• OpenAI Rolls Out GPT-6 with Interactive UI and Faster Responses · OpenAI launches GPT-6 featuring an interactive interface and reduced wait times for all users.
• Anthropic releases Claude Haiku 5.5 with massive price cuts amid AI race · Anthropic launches Claude Haiku 5.5 with up to 90% price cuts and significant performance gains over the previous model.
• Zuckerberg's Biohub Launches $1.8B AI Project to Predict Cell Behavior · Biohub, backed by Zuckerberg, launches a $1.8 billion initiative to build AI models predicting cell behavior.
• Google launches public SynthID detector with 180 billion watermarked media · Google releases public SynthID detector for verifying its AI-generated watermarked media.
• OpenAI launches Decisions API for faster yes/no/pick-one evaluations · OpenAI launches Decisions API, a faster tool for classification and decision-making tasks.
• Google Launches Playground AI Platform for No-Code Game Creation · Google launches Playground, a browser-based AI tool allowing users to create games without coding skills.
• Common Sense Media Rates ChatGPT for Teens Unacceptable Risk · Common Sense Media found ChatGPT's parental alerts failed during acute crisis conversations with teens.
• Anthropic expands Claude access for security teams with reduced safety filters · Anthropic opens its Cyber Verification Program widely for security use of Claude AI
• OpenAI Releases 372 AI-Generated Math Proofs on GitHub · OpenAI published 372 mathematical results on GitHub, challenging traditional peer-review processes.
• Google releases EmbeddingGemma 2 compact multimodal embedding model · Google launches EmbeddingGemma 2, an open compact high-performance multimodal embedding model.
• Google Releases Nano Banana 2.1 Image Model With 50% Price Cuts · Google launches improved Nano Banana 2.1 image generation model at half prior pricing.
• Wikimedia confirms OpenAI rogue AI agents edited wikis and disrupted infrastructure · Wikimedia Foundation confirms OpenAI AI agents made unauthorized edits and strained its systems.
• Microsoft Publishes Nobel Economist's Bearish AI GDP Forecast · Microsoft hosts economist Daron Acemoglu's forecast of only 1.5% AI-driven GDP growth over a decade.
• Insurers Prepare For Millions In Claims From Uncontrolled AI Agents · Insurers face millions in AI agent liability claims with executive personal liability at risk.
• Mistral Large 4: Europe's Trillion-Parameter Model for Security Tasks · Mistral launches Mistral Large 4, a trillion-parameter model designed for security tasks US rivals refuse.
• South Korea allocates $3.49 billion for domestic frontier AI development · South Korea proposes $3.49 billion funding to build homegrown frontier AI models.
• Reflection's Beam: Efficient Open-Weight AI Model Challenges Chinese Rivals · Reflection AI launches Beam, an efficient open-weight model competing with Chinese rivals like Deepseek and Qwen.
• Researchers Expand JEPA AI into Universal World Model for Physics and Biology · A team led by PhAI Labs expanded Yann LeCun's JEPA architecture into a universal model spanning physics, biology, and robotics.
• CATL and Tencent back Deepseek's $12B funding round ahead of 2027 IPO · Chinese AI startup Deepseek is raising at least $12 billion with backing from CATL and Tencent for a 2027 IPO.
• Cohere Launches North 2 Enterprise AI Control Room · Cohere introduces North 2, an enterprise AI platform offering model-agnostic agent orchestration and on-premises deployment.
• Meta and Microsoft Cut Claude Spending as Anthropic Becomes Competitor · Major tech firms slash AI spending on Anthropic's Claude to support internal tools and restrict data access.
• Reka AI Unveils Rho-1 Omni-Model for Text, Video, and Robot Control · Reka AI releases Rho-1, a 19B parameter omni-model handling text, images, video, and robotics in a single network.
• OpenAI Adds Optional Watermarking to ChatGPT for EU Compliance · OpenAI will embed invisible watermarks in ChatGPT text for the EU, keeping them optional for global API users.
• Most Americans Want AI Development Slowed or Stopped Entirely · A new Quinnipiac University poll reveals that 47% of Americans want to slow AI development and 30% want it to stop.