News and analysis on Generative AI, each with sources, continuously updated from our radar and research.
OpenAI is reportedly planning to directly purchase Google's TPU chips. The move aims to decrease its dependence on Nvidia GPUs and cut costs for running frontier AI models.
Apple has updated Final Cut Pro, Pixelmator Pro, and Logic Pro with generative AI tools, adding native rotoscoping, text-driven inpainting, and enhanced drum stem splitting.
A 67-page paper by OpenAI sparks debate after an internal AI model claimed to disprove the Exponential Time Hypothesis on the 3-SAT problem. Review with the original authors is underway.
Legora uses OpenAI models, ChatGPT, and Codex to accelerate workflows and expand contract analysis, allowing legal professionals to focus on higher-value tasks.
Legora is utilizing OpenAI models, ChatGPT, and Codex to speed up legal workflows and broaden contract analysis, according to OpenAI.
Google DeepMind has titled a blog entry Gemini 4 Carbon, identifying the project as its next frontier model.
NVIDIA has presented Nemotron 3.5 Transcribe and Granary v2, highlighting multilingual speech transcription that users can prompt and adapt.
Internal AI models at OpenAI and Anthropic reportedly found a subexponential algorithm for 3-SAT, challenging a 25-year-old pillar of complexity theory. An expert panel is now auditing the proof.
OpenAI and ServiceNow are expanding their partnership to integrate OpenAI models and agents into ServiceNow, while bringing ServiceNow workflows into ChatGPT and Codex.
AWS has published a blueprint for building a governed, multi-account agentic AI platform using Amazon Bedrock AgentCore, featuring isolated workstream cells and automated evaluation gates.
Anthropic officially filed its S-1 prospectus to list under ticker ANT, disclosing $16.3 billion in annualized revenue run-rate alongside significant copyright liabilities.
Following the rise of AI models, decomp.dev has restricted core features as automated reverse-engineering divides the modding community and sparks licensing disputes.
Anthropic has filed its S-1 registration statement to list on Nasdaq under ANT, revealing massive revenue growth, substantial cloud commitments and extensive risk warnings.
OpenAI has launched GPT-6.1 Terra, a mid-tier model designed for computer use. It cuts token prices in half and executes agent tasks three times faster than previous flagship models.
OpenAI and Oracle are expanding their partnership to bring business applications and AI agents into ChatGPT and power new Oracle offerings.
Check Point has integrated OpenAI Daybreak to speed up research into new security protections and convert vulnerability intelligence into customer defenses faster.
Google DeepMind has announced Genie 4, describing the system as a new world model designed to unlock unprecedented realism in real-time simulation.
Teaming up with Google Gemini and UnternehmerTUM, FC Bayern Munich has launched the Fan AI-DEATION competition to co-create generative AI features with fans.
Between high-tech speed and fan anger: While Sportel Monaco showcased live AI tools, FOX Sports triggered viewer backlash by using AI-generated manager portraits during the World Series.
Developer Simon Willison demonstrates how Claude Opus 5.5 builds multiplayer browser games and standalone canvas engines entirely through targeted prompting.
The Elvis Presley estate and Reservoir Media are suing Anthropic and Suno for hundreds of millions of dollars over alleged copyright infringement in AI training.
Scuderia Ferrari HP and IBM have introduced AI agents to the official team app, converting live telemetry and team radio into detailed race summaries within minutes.
Moonshot AI announces Kimi K3.1 Pro: The MoE model features 2.8 trillion parameters and upcoming open weights in November, while safety filtering curbs offensive cyber exploits.
Epic Games launches the MegaJam 2026 with substantial cash prizes, but removes mandatory generative AI requirements following sharp pushback from the gaming developer community.
Actor unions in Japan and the UK are pushing back against unauthorized AI voice cloning and buyout contracts that surrender vocal rights to train generative speech models.
A benchmark report by Kahua and Verdantix reveals that while 84 percent of construction firms use AI, only eight percent achieve significant ROI due to severe system fragmentation.
Tennis Australia has named Adobe its official AI partner for the Australian Open, leveraging generative personalization for on-site fans at Melbourne Park and global audiences.
Moonshot AI has unveiled Kimi K3.1 Pro, a 2.8-trillion-parameter MoE model with a one-million-token context window designed for financial analysis and agentic workflows.
OpenAI is reportedly closing a 30 billion dollar funding round at a 1.25 trillion dollar valuation, structured by Coatue and UAE investors as a bridge to a planned 2027 IPO.
With its 'Human Made, Human Paid' campaign, the AFM union demands consent and compensation from Universal, Sony, and Warner regarding commercial AI licensing deals.
AI music platform Suno launched a Live Experiences division led by former Ticketmaster VP Michael Chua, integrating Songkick to bring computer-generated music to physical stages.
Moonshot AI has announced Kimi K3.1 Pro, an open-weight mixture-of-experts model with 2.8 trillion parameters tailored directly for game engines and parametric 3D pipelines.
Former RE/MAX CEO Nick Bailey has joined AuthoritySignal AI as a founding member to help real estate brokers optimize their visibility across generative search engines.
The University of Texas has partnered with tech giants to create Bevo Lab, deploying generative AI assistants and analytics across 21 collegiate teams and over 500 student-athletes.
New studies from Ricoh and Founder Reports show workplace AI tools often cause friction and extra work, as software rollouts outpace change management and training.
At the Berlin investment summit, the German government and Nvidia signed a memorandum of understanding to advance industrial AI foundation models for manufacturing and mid-sized enterprises.
Universal Music Group and TYGER have struck a licensing deal enabling AI models to dynamically reshape catalog tracks in real time based on active gameplay.
Ahead of the 2026/27 season, the NBA launches generative tool 'Ask NBA' and 'Game Pulse' featuring new AWS-powered live metrics.
Financial institutions and payment providers at GTC Berlin are presenting production AI systems built on open models, while Revolut adopts its PRAGMA model on NVIDIA infrastructure.
New empirical data reveals European developers shifting heavily toward Chinese models, even as real-world productivity increases remain modest and governance lags behind.
Replit has replaced Claude as the default model in Replit Agent with OpenAI to curb inference costs, as DeepSeek launches V4.1-Flash and pressure mounts across developer platforms.
MiniMax has introduced its M3.1 Flash MoE model, combining low inference costs with strong benchmark scores to target automated banking and compliance workflows.
SF Compute CEO Evan Conrad outlined an order book for compute power on Latent Space. Physical settlement is intended to curb arbitrage between raw GPU contracts and B300 inference revenues.
A new study by Anthropic reveals a sharp divide: while AI insiders expect massive automation by 2030, the general public anticipates a far slower transformation of everyday work.
Recent Bain & Company studies reveal a growing disconnect between accelerated code generation, surging operational costs, and missed enterprise return-on-investment targets.
