Recent Posts
Deals•Jan 30, 2026
OpenAI Announces Plans for Late‑2026 IPO
OpenAI disclosed that it is preparing for an initial public offering in the fourth quarter of 2026, targeting a valuation of $830 billion and a raise of over $100 billion. The startup is in informal talks with Wall Street banks and has hired senior finance staff. SoftBank and Amazon are reported to be considering large investments ahead of the IPO.
THE DECODER
News•Jan 27, 2026
Mistral AI Launches Terminal-Based Coding Agent Vibe 2.0
Mistral AI introduced Vibe 2.0, a terminal‑based coding agent built on the new Devstral 2 model. The assistant lets developers command codebases with natural language, manage multiple files at once, and retain full project context. Version 2.0 adds custom subagents for testing and...
By THE DECODER
News•Jan 27, 2026
Anthropic CEO Warns Democracies Must Protect Themselves From Their Own AI
Anthropic CEO Dario Amodei released an essay warning that democracies must use AI responsibly, avoiding tools that make them resemble autocratic regimes. He draws hard lines against AI‑powered domestic mass surveillance and long‑term personalized propaganda, while endorsing limited use of...
By THE DECODER
News•Jan 26, 2026
MCP Apps, the Model Context Protocol's First Official Extension, Turns AI Responses Into Interactive Interfaces
The episode explains the launch of MCP Apps, the first official extension to the Model Context Protocol, which lets AI tools embed interactive dashboards, forms, visualizations, and multi‑step workflows directly into chat conversations. It highlights how this solves the problem...
By THE DECODER
News•Jan 26, 2026
Emergency Meetings and Failed Billion-Dollar Talks Reveal the Chaos Behind Apple's Pivot to Google Gemini
The episode unpacks Apple’s turbulent shift from building its own AI models to partnering with Google’s Gemini, sparked by a Bloomberg leak that prompted an emergency meeting where Siri chief Mike Rockwell dismissed the report as "bullshit." Internal strife, talent...
By THE DECODER
Podcast•Jan 26, 2026•0 min
Nvidia Releases Open Model PersonaPlex, a Voice AI that Listens and Talks at the Same Time
Nvidia unveiled PersonaPlex, an open‑source 7 B parameter voice AI that can listen and speak simultaneously, delivering real‑time, natural conversations with customizable voices and roles via hybrid audio‑text prompts. By combining real and synthetic training data, the model learns nuanced behaviors...
By THE DECODER
News•Jan 25, 2026
ChatGPT Fails to Spot 92% of Fake Videos Made by OpenAI's Own Sora Tool
Newsguard tested three leading chatbots on 20 AI‑generated videos created by OpenAI’s Sora tool and found that ChatGPT misidentified 92.5% of them as authentic. xAI’s Grok performed slightly worse at 95% error, while Google’s Gemini was the most accurate, correctly...
By THE DECODER
News•Jan 25, 2026
Vibe Coding May Be Behind a 60 Percent Spike in New iOS Apps
Data from Sensor Tower and Wells Fargo Securities shows a 60 percent year‑over‑year jump in new iOS apps in December 2025, with overall 2025 growth of 24 percent. The spike aligns with the rollout of agentic coding tools—AI‑powered platforms that let users create...
By THE DECODER
News•Jan 24, 2026
Apple Study Reveals AI Controllability Is Fragile and Varies Wildly by Task and Model
Apple researchers and Universitat Pompeu Fabra introduced a control‑theoretic framework that measures how reliably generative models follow user instructions. Tests on language models such as Gemma‑3‑B, Qwen‑3‑4B and SmolLM‑3‑B revealed stark differences in controllability and calibration across simple tasks like...
By THE DECODER
News•Jan 24, 2026
OpenAI's GPT-5.2 Pro Solves Math Problems that Stumped Every AI Model Before It
OpenAI unveiled GPT-5.2 Pro, which achieved a 31 percent success rate on the toughest FrontierMath Tier 4 benchmark, eclipsing Gemini 3 Pro’s 19 percent. The model solved 15 of 48 tasks, including four problems no AI had previously cracked. Mathematicians praised the breakthroughs but flagged...
By THE DECODER
News•Jan 24, 2026
Google Deepmind's D4RT Model Aims to Give Robots and AR Devices More Human-Like Spatial Awareness
Google DeepMind unveiled D4RT, a Dynamic 4D Reconstruction and Tracking model that builds a unified scene representation from video and answers spatiotemporal queries with a single lightweight decoder. The architecture collapses depth estimation, motion tracking, and camera pose into one...
By THE DECODER
News•Jan 24, 2026
One in Ten Japanese Creatives Loses Income to Generative AI
A Freelance League of Japan survey of 24,991 creators shows 12 percent of manga artists, illustrators and other freelancers experienced income declines due to generative AI. Nine‑point‑three percent lost between 10 and 50 percent of earnings, while 2.7 percent saw more than half disappear. Clients...
By THE DECODER
News•Jan 24, 2026
Anthropic Opens Claude's Improved Excel Integration to All Pro Subscribers After Limited Beta
Anthropic has expanded its Claude‑powered Excel add‑in from a limited beta to all Pro subscribers as of Jan 24, 2026. The rollout adds multi‑file drag‑and‑drop, safeguards against overwriting cells, and introduces automatic session compression. These enhancements build on earlier financial‑analyst tools, including...
By THE DECODER
News•Jan 24, 2026
OpenAI Developer Predicts Programmers Will Soon "Declare Bankruptcy" On Understanding Their Own AI-Generated Code
OpenAI developer known as “roon” warns that programmers will soon openly admit they no longer fully understand the AI‑generated code they commit, a phenomenon he calls “declaring bankruptcy.” A recent 2025 developer survey shows only 33% trust AI‑written code, yet...
By THE DECODER
News•Jan 24, 2026
Microsoft-Tsinghua Team Trains 7B Coding Model that Beats 14B Rivals Using only Synthetic Data
Researchers from Microsoft and Tsinghua unveiled SynthSmith, a pipeline that creates fully synthetic coding tasks, solutions, and tests. Using this data, the 7‑billion‑parameter X‑Coder model achieved a 62.9% pass rate on LiveCodeBench v5 and 55.8% on v6, beating 14‑billion‑parameter rivals. Experiments...
By THE DECODER
News•Jan 23, 2026
OpenAI's Upcoming Codex Update Will Hit the Company's "High" Cybersecurity Risk Level for the First Time
OpenAI announced that upcoming updates to its Codex code‑generation model will push it into the “High” cybersecurity risk tier in the company’s internal risk framework, the first time a model has reached that level. The “High” designation means Codex could...
By THE DECODER
News•Jan 22, 2026
OpenAI CMO Responds to "Woke AI" Accusations by Citing Co-Founder Brockman's $25 Million MAGA Donation
OpenAI’s chief marketing officer, Kate Rouch, pushed back against accusations that the company is a "Woke AI" by highlighting her own Republican background and a $25 million MAGA donation from co‑founder Greg Brockman. The criticism originated from MAGA activist Katie Miller,...
By THE DECODER
News•Jan 22, 2026
Elon Musk's AI Chatbot Grok Flooded X with Millions of Sexualized Images
Elon Musk’s X‑integrated AI chatbot Grok generated at least 1.8 million sexualized images in a nine‑day span, according to analysis by the New York Times and the Center for Countering Digital Hate. Roughly 65 % of a 20,000‑image sample depicted sexualized adults or children,...
By THE DECODER
News•Jan 22, 2026
Ernie 5.0: Baidu's 2.4 Trillion Parameter Model Becomes China's Best in LMArena
Baidu has launched Ernie 5.0, a 2.4‑trillion‑parameter multimodal model that handles text, images, audio and video through a unified architecture. In the LMArena benchmark released Jan 15 2026, Ernie‑5.0‑0110 earned 1,460 points, ranking eighth worldwide and first among Chinese models, matching OpenAI’s GPT‑5.1...
By THE DECODER
News•Jan 21, 2026
Ollama Brings Local AI Image Generation to Mac
Ollama, the platform for running AI models locally, has launched an experimental image‑generation feature for macOS. The initial release includes two models: Alibaba’s Z‑Image Turbo, a 6‑billion‑parameter photorealistic generator, and Black Forest Labs’ Flux 2 Klein, offered in 4‑billion and 9‑billion‑parameter variants....
By THE DECODER
News•Jan 21, 2026
OpenAI Promises AI Data Centers Won't Raise Local Electricity Prices
OpenAI announced its Stargate Community program, pledging that new AI data centers will not raise local electricity rates. The company will finance its own power generation, battery storage, and grid upgrades tailored to each site, while also protecting water resources...
By THE DECODER
News•Jan 21, 2026
Interactive Demo Shows AI Models Have Opinions - and Grok Really Likes Elon Musk
The nonprofit CivAI released an interactive demo that asks twenty leading AI models a series of ethical, social and political questions, allowing users to compare model answers with their own. The results show stark differences, with most models favoring Mahatma...
By THE DECODER
News•Jan 20, 2026
OpenAI Rolls Out Age Prediction to Apply Teen Safeguards in ChatGPT
OpenAI is introducing an age‑prediction feature in ChatGPT that automatically identifies accounts likely belonging to users under 18. When a minor is flagged, the system activates a safety mode that blocks graphic violence, sexual role‑play, self‑harm content, and extreme beauty...
By THE DECODER
News•Jan 20, 2026
Deepmind and Anthropic CEOs Expect AI to Hit Entry-Level Jobs and Internships in 2026
DeepMind and Anthropic CEOs warn AI will affect entry‑level jobs by 2026. Hassabis observed early impact at DeepMind, predicting internships could shrink this year. Amodei forecasts up to half of office jobs for young professionals disappearing within five years as...
By THE DECODER
News•Jan 20, 2026
Software Stocks See Worst Start to Year Since 2022 as AI Disruption Fears Mount
Anthropic unveiled Claude Cowork, an AI‑powered office assistant, sparking fresh fears about software disruption. The announcement triggered a sharp sell‑off in software equities, with Intuit down 16%, Adobe and Salesforce each losing over 11%, marking the worst weekly decline for the...
By THE DECODER
News•Jan 19, 2026
Google's Gemini API Requests More than Double in Five Months, Jumping From 35 Billion to 85 Billion
Google’s Gemini API requests more than doubled, climbing from roughly 35 billion in March to about 85 billion by August. The surge followed the launch of Gemini 2.5, which turned profitable on operating costs, and continued with the rollout of Gemini 3. Gemini Enterprise...
By THE DECODER
News•Jan 19, 2026
Former OpenAI Policy Chief Launches Institute for Independent AI Safety Audits
Miles Brundage, former OpenAI policy chief, launched the AI Verification and Evaluation Research Institute (AVERI) to champion independent safety audits of frontier AI models. AVERI has raised $7.5 million and aims for $13 million to support a 14‑person team, with funding from...
By THE DECODER
News•Jan 18, 2026
South Koreans Now Spend More on AI Subscriptions than Netflix Each Month
South Korean consumers and businesses spent an estimated 80.3 billion won on AI subscriptions in December 2025, surpassing the average monthly Netflix revenue of 75 billion won in 2024. Credit‑card transactions for AI services surged from 52,000 in January 2024 to 1.666 million by year‑end,...
By THE DECODER
News•Jan 18, 2026
Snap's SnapGen++ Generates High-Resolution AI Images on iPhone in Under Two Seconds
Snap Inc. unveiled SnapGen++, a 0.4 billion‑parameter diffusion‑transformer that creates 1024 × 1024‑pixel images on an iPhone 16 Pro Max in just 1.8 seconds. The model replaces the earlier U‑Net‑based SnapGen and introduces a novel attention scheme that slashes per‑step latency from 2,000 ms to under 300 ms. Using...
By THE DECODER
News•Jan 18, 2026
Google Upgrades AI Overviews with Gemini 3 Pro for Complex Queries
Google is integrating its newest Gemini 3 Pro model into AI Overviews, the concise answer boxes that appear beneath search results. The system now intelligently routes complex queries to Gemini 3 Pro while delegating simpler questions to faster, less expensive...
By THE DECODER
News•Jan 18, 2026
Even the Best AI Models Fail at Visual Tasks Toddlers Handle Easily
A new BabyVision benchmark reveals that leading multimodal AI models struggle with basic visual tasks that toddlers master before speaking. The top model, Gemini‑3‑Pro‑Preview, achieved only 49.7% accuracy, far below the 94.1% score of adult humans and the average three‑year‑old...
By THE DECODER
News•Jan 17, 2026
Elon Musk Proposed His Children Should Inherit Control of Human-Like AI, OpenAI Claims
Elon Musk reportedly told OpenAI officials that his children should inherit control of artificial general intelligence (AGI) systems. OpenAI alleges Musk’s push for a majority stake in the company was tied to this succession plan, echoing his earlier ambition to...
By THE DECODER
News•Jan 16, 2026
OpenAI Wants Its API Format to Become the Industry Standard
OpenAI has introduced "Open Responses," an open‑source interface that standardizes request and response formats across language models from multiple providers. The initiative builds on OpenAI’s Responses API and aims to let developers write code once and run it on any...
By THE DECODER
News•Jan 16, 2026
OpenAI Will Soon Test Ads in ChatGPT Despite CEO Sam Altman Once Calling the Idea Dystopian
OpenAI announced it will begin testing advertisements in ChatGPT for logged‑in adult users in the United States within weeks. The ads will appear only on the free tier and the newly expanded $8‑per‑month ChatGPT Go subscription, while Pro, Business and Enterprise...
By THE DECODER
News•Jan 16, 2026
Study Warns AI Could Homogenize Human Creativity as Models Converge on "Artificial Hivemind"
Researchers from the University of Washington, Carnegie Mellon and the Allen Institute for AI found that a wide range of large‑language models produce nearly identical responses to open‑ended prompts, a phenomenon they label the “Artificial Hivemind.” In tests with 25...
By THE DECODER
News•Jan 16, 2026
OpenAI Safety Researcher Joins Anthropic's Alignment Team
Senior safety researcher Andrea Vallone has left OpenAI to join Anthropic’s alignment team. At OpenAI she founded the Model Policy research group and contributed to GPT‑4, GPT‑5, and advanced reasoning models. Her recent work tackled how AI should respond to...
By THE DECODER
News•Jan 15, 2026
Google's New Open TranslateGemma Models Bring Translation for 55 Languages to Laptops and Phones
Google unveiled TranslateGemma, an open‑weights translation suite available in 4‑B, 12‑B, and 27‑B versions for mobile, laptop, and cloud use. The 12‑B model surprisingly beats the larger 27‑B baseline, achieving a MetricX error score of 3.60 versus 4.04, a 26%...
By THE DECODER
News•Jan 15, 2026
OpenAI Launches Call for US-Based AI Hardware Suppliers in Push for Domestic Manufacturing
OpenAI has issued a request for proposals to source U.S.-based manufacturers of AI data‑center hardware, including cooling, power, networking, consumer electronics, and robotics components. The program, open through June 2026, aims to build a domestic supply chain that mirrors the broader...
By THE DECODER
News•Jan 15, 2026
Anthropic Cuts AI Productivity Forecasts in Half After Analyzing Claude's Real-World Failure Rates
Anthropic’s fourth Economic Index Report reveals Claude’s success rate falls sharply as task complexity rises, with roughly 60% success on sub‑hour API requests and 45% on tasks exceeding five hours. The new data forced the company to halve its AI‑driven...
By THE DECODER
News•Jan 14, 2026
OpenAI Opens GPT-5.2 Codex to Developers Through the Responses API
OpenAI has launched GPT-5.2 Codex through its Responses API, moving the model out of the restricted Codex environment. The multi‑modal model accepts text and images and offers four configurable reasoning levels, targeting complex development tasks such as feature creation, refactoring,...
By THE DECODER
News•Jan 14, 2026
Anthropic's Labs Team Gets a Shake-Up as Instagram Co-Founder Mike Krieger Joins Experimental AI Unit
Anthropic is expanding its Labs unit, the experimental arm that builds Claude AI products, by moving former Chief Product Officer and Instagram co‑founder Mike Krieger into the team with Ben Mann. New product lead Ami Vora, who joined in late...
By THE DECODER
News•Jan 14, 2026
Despite OpenAI Partnership, Microsoft Is One of Anthropic's Biggest Customers
Microsoft is on track to spend nearly $500 million a year on Anthropic’s AI models, integrating them into products such as GitHub Copilot and Office 365. The spend follows a up‑to‑$5 billion investment announced last November, positioning Anthropic as a key AI...
By THE DECODER
News•Jan 14, 2026
Deutsche Telekom Puts Elevenlabs AI on the Phone to Handle Customer Calls
Deutsche Telekom will integrate Elevenlabs’ AI voice agents into its customer service channels, allowing callers to interact with realistic synthetic voices 24/7 via the MeinMagenta app and phone. The rollout follows a pilot that lets Magenta customers generate up to...
By THE DECODER
News•Jan 14, 2026
New Apple-Google Deal Pushes ChatGPT to the Sidelines on iPhone
Apple has deepened its partnership with Google by embedding the Gemini large‑language model directly into Siri, making it the default AI for routine iPhone queries. The company confirmed that its existing agreement with OpenAI will keep ChatGPT available for more...
By THE DECODER
News•Jan 14, 2026
Bandcamp Bans AI-Generated Music
Bandcamp announced a new policy that bans music created entirely or substantially by generative AI, including tracks that imitate specific artists or styles. The platform, which emphasizes direct purchases and artist‑fan relationships, will allow users to report suspected AI‑generated content...
By THE DECODER
News•Jan 13, 2026
AI Images Complicate Search for Escaped Monkeys in St. Louis
Several vervet monkeys escaped in St. Louis, prompting a citywide search. Social media is flooded with AI‑generated images falsely claiming the animals have been captured, confusing the public and officials. The health department warns residents to keep distance as the...
By THE DECODER
News•Jan 13, 2026
AI Models Don't Have a Unified "Self" - and That's Not a Bug
Anthropic researcher Josh Batson argues that language models like Claude do not maintain a unified internal representation of facts. Experiments show the model uses separate mechanisms to retrieve knowledge and to verify statements, leading to contradictory answers. This internal fragmentation...
By THE DECODER
News•Jan 13, 2026
Anthropic's Claude Cowork Was Built in Under Two Weeks Using Claude Code to Write the Code
Anthropic unveiled Claude Cowork, an AI‑powered collaboration app built in roughly one and a half weeks. The majority of its code was generated by Claude Code, the company’s own code‑generation model, according to inventor Boris Cherny. Product manager Felix Rieseberg said the sprint‑and‑a‑half effort...
By THE DECODER
News•Jan 11, 2026
Chinese Researchers Diagnose AI Image Models with Aphasia-Like Disorder, Develop Self-Healing Framework
Chinese researchers introduced UniCorn, a self‑healing framework that teaches multimodal AI models to recognize and correct mismatches between image understanding and generation, a problem they label “Conduction Aphasia.” The system splits a single model into Proposer, Solver, and Judge agents...
By THE DECODER
News•Jan 11, 2026
Lightricks Open-Sources AI Video Model LTX-2, Challenges Sora and Veo
Lightricks has open‑sourced its 19‑billion‑parameter LTX‑2 model, which generates synchronized stereo audio and video from a single text prompt. The dual‑stream transformer allocates 14 B parameters to video and 5 B to audio, enabling 4K, 50 fps output up to 20 seconds long. Benchmarks...
By THE DECODER
News•Jan 11, 2026
Chinese AI Industry Admits US Remains Ahead for Now
Chinese AI leaders acknowledge the United States remains ahead in compute power and chip access, limiting near‑term chances of overtaking OpenAI or Anthropic. Executives at the AGI Next Summit cited export controls and a two‑order‑of‑magnitude gap in infrastructure as primary...
By THE DECODER
Deals•Jan 1, 2026
Baidu's Kunlunxin Files Confidential Hong Kong IPO, Valued at $3B
Baidu's AI chip division Kunlunxin has confidentially filed for an IPO in Hong Kong, submitting its application on Jan 1. A recent financing round values the unit at roughly $3 billion, though the final offering size remains undetermined. The filing adds Kunlunxin...
THE DECODER
Deals•Jan 1, 2026
Moonshot AI Secures $500M Series C Funding Led by IDG
Chinese AI startup Moonshot AI announced a $500 million Series C round that values the company at $4.3 billion. The round was led by IDG with $150 million and included Alibaba, Tencent and individual investor Wang Huiwen. The capital will fund Kimi‑K3 development and...
THE DECODER