π°
New research enables one-shot long-horizon parsing
This development could improve AI's ability to understand and process complex data over extended periods.
π€ AI β 2026-07-21
ddNews Β· a daily AI briefing for Dirk (generative video/ddStudio, Lit
π°
Bifrost AI gateway unifies access to 23+ providers
The platform offers a single API for models like OpenAI and Anthropic.
Chinese tech giants are slashing LLM prices
iFlytek is offering SparkDesk LLM for free or at significantly reduced costs.
π°
Why is reading good for me?
A general interest article on the benefits of reading.
EU launches AI Hub, no disclosure rules
The platform offers access to models from major AI companies.
π°
Poolside releases Laguna S 2.1
This open-weight coding model matches larger models and runs on a single desktop.
Azure challenges Nvidia with AMD Helios
Microsoft is deploying AMDβs open-standard rack to reduce reliance on Nvidiaβs infrastructure.
EU launches AI hub, lacks disclosure rules
The EPGenAI Hub provides access to models from major AI companies, but lawmakers aren't required to disclose AI drafting.
Geopolitics of open-weight models
The increasing capability of open-weight models raises concerns about geopolitical implications.
π°
University of Tennessee sues Anthropic
UT claims Anthropic infringed on its patents.
Anthropic settles copyright lawsuit for $1.5B
The settlement addresses claims of using pirated books to train Claude.
Cisco launches low-cost AI models for code security
Cisco Foundation AIβs Antares is a small language model designed for vulnerability hunting.
Samsung establishes new robotics business
The company sees physical AI as its next frontier.
Kimi K3's global AI impact assessed
The model is challenging the US duopoly in frontier AI.
Anthropic pays $1.5B in Claude copyright settlement
A federal judge approved the settlement.
Anthropic pays authors $1.5B in settlement
The settlement resolves copyright issues related to Claude AI training.
Google releases Gemini 3.6 Flash and Cyber
The company expands its Gemini lineup with cheaper, more efficient models.
Google delays Gemini 3.5 Pro
The continued absence of 3.5 Pro raises questions about Googleβs AI strategy.
American startups aim to beat Chinese AI
Thinking Machines and others are challenging Kimi and Deepseek.
π°
Sonilo and fal Launch Sound Effects 1.0
Sonilo and fal launched a tool for realistic sound effects from video and text.
π°
Anthropic settlement approved
The agreement is the largest known copyright recovery in history.
Judge approves Anthropic settlement
The $1.5B settlement with authors is officially approved.
New York school introduces AI robot teacher Sally
Salamanca High School is using a 5'2" humanoid robot as a teaching assistant.
Samsung unveils new robotics business
Samsung is establishing a new unit focused on physical AI and robotics.
Google develops custom AI chip
Google is building a chip to run Gemini models more efficiently.
Google expands Gemini Flash lineup
New models aim to make AI agent workflows faster and cheaper.
US judge approves Anthropic settlement
The landmark settlement involves pirated books used for training.
Anthropic pays $1.5B in copyright settlement
Anthropic settled a copyright lawsuit over data used to train Claude.
Anthropic settles copyright lawsuit
A $1.5 billion settlement has been approved for authors whose books were used to train Claude.
Microsoft expands Mistral AI partnership
A multibillion-dollar commitment will bring LLMs to Europe.
Google launches Mythos rival
Google expands Gemini with cheaper models and a new cybersecurity offering.
Google doubles down on cheaper AI
Google unveiled Gemini 3.6 Flash and 3.5 Flash Lite, focusing on cost reduction.
Judge approves Anthropic settlement
A federal judge approved the $1.5 billion copyright settlement.
π°
Google releases cheaper Gemini models
Google launched Gemini 3.6 Flash, 3.5 Flash Lite, and Flash Cyber, focusing on cost reduction.
Google updates lightweight Gemini models
Google released cheaper Gemini versions, but the flagship model remains delayed.
Google updates lightweight Gemini models
Google released cheaper versions of Gemini, but the flagship model remains delayed.
Google expands Gemini with cheaper models
Google is releasing more efficient models and a new cybersecurity offering.
Claude Fable cracks a math problem
Claude Fable 5 reportedly disproved the Jacobian Conjecture, demonstrating AIβs growing mathematical capabilities.
Anthropic settles copyright lawsuit for $1.5B
Anthropic will pay $1.5 billion over pirated books used for training, with implications for other AI models.
Anthropic settlement approved
A $1.5 billion copyright settlement involving Anthropic has been finalized.
10 million users on OpenAI's Codex and ChatGPT Work
OpenAI reports significant adoption of its agent products.
AMD sees open-source AI as a catalyst
AMD is positioned to benefit from the growth of open-source AI workloads.
Googleβs new chip unlocks AI backlog
The chip is key to addressing the $462 billion demand for AI computing power.
π°
Infinity aims to break Nvidiaβs AI chip grip
The startup raised $15M to build software that makes any AI chip run models faster.
π°
Poolside releases Laguna S 2.1
The open-weight model matches larger models on coding tasks.
π°
AI breaks 87-year-old maths problem
Anthropicβs Fable 5 found a counterexample to the Jacobian conjecture.
π°
Microsoft & Mistral expand AI partnership
The collaboration aims to provide enterprises and regulated industries with more control over frontier AI.
Michaels launches 'Ask Mike' AI assistant
The assistant is built with Google Cloud's Gemini Enterprise for customer experience.
π°
Chinese AI agent outperforms Claude Code
Zhejiang Universityβs Qiushi Engine topped a research benchmark.
Chinese AI agent outperforms Anthropic's Claude Code
Zhejiang University's Qiushi Engine topped a research benchmark.
Alibaba's Qwen3.8-Max claims second place
The 2.4-trillion-parameter model lacks published benchmarks.
Anthropic to pay β¬1.3bn in copyright settlement
The payout closes the first major AI copyright case.
Z.ai GLM 5.2 is used to combat cyberattacks
Hugging Face utilized a Chinese model due to limitations with US guardrails.
π°
Chinese AI agent outperforms Claude Code
Zhejiang Universityβs Qiushi Engine leads the ResearchClawBench leaderboard.
Chinaβs Z.ai exposes coder habits
Low-priced Chinese models reveal costly token usage patterns.
Anthropic $1.5 billion settlement approved
The payout covers an estimated 500,000 titles.
Anthropic copyright settlement details
The agreement includes a $3,000 payout per work, but lacks binding precedent.
Google brings air-gapped Gemini to India
Google offers a secure, data-resident Gemini option for regulated Indian enterprises.
US needs to stop being surprised by Chinese AI
Experts argue the US should anticipate advancements from China.
π°
Faceless YouTube channels are growing faster
The trend is attributed to various factors.
Anthropic settles copyright lawsuit for $1.5B
The settlement covers books used to train Claude.
Google's Frozen v2 chip hardwires Gemini
The chip aims for six to ten times the inference efficiency of current solutions.
Claude Code seals Bash/Unicode bypasses
Update v2.1.216 addresses permission issues in auto mode.
Anthropic settles copyright lawsuit for $1.5B
The settlement involves pirated books used for Claude training.
LSE Launches 24/5 AI Trading Venue
The London Stock Exchange is building an AI-native trading platform with on-chain settlement.
Kimi K3 outperforms Claude, but hallucinates
The model tops coding leaderboards but suffers from a 51% hallucination rate.
Anthropic settles copyright lawsuit for $1.5B
The settlement resolves claims of using pirated books for training.
Google updates lightweight Gemini models
Google released cheaper versions of Gemini, but the flagship model remains delayed.
Google announces Gemini 3.6 Flash & cybersecurity AI
Google unveiled Gemini 3.6 Flash and teased Gemini 3.5 Pro, focusing on cost reduction.
Anthropic settles copyright lawsuit for $1.5B
A federal judge approved the record-breaking settlement with authors.
Google DeepMind's India Strategy
Google is focusing on making AI cheaper, safer, and better at coding.
π°
Novo Nordisk appoints new Gulf chief
The appointment reflects regional growth and innovation.
Canadaβs new AI data centres may be obsolete
The growth of small language models could render them unnecessary.
Kimi K3 Exposes Open-Weights Costs
Moonshot AI's Kimi K3 highlights the infrastructure demands of large open-source models.
Anthropic settles copyright lawsuit for $1.5B
A US judge approves Anthropicβs $1.5 billion settlement in a copyright lawsuit over misuse of authors' works.
Sonilo and fal launch Sound Effects 1.0
A new API co-launch brings Video-to-Sound Effects and Text-to-Sound Effects generation to developers.
Microsoft deploys AMD Helios AI systems
Azure will add AMD EPYC VMs and Pensando networking.
Anthropic offers AI credits for rare disease research
The initiative supports researchers using AI to find treatments.
Chip stock recommendations
Experts suggest buying Samsung and selling Hynix.
Kimi K3 fuels US AI safety concerns
A new Chinese model is approaching OpenAI's capabilities.
Google expands Agentic AI to university
Gemini Enterprise rolls out to 10,000 students and staff.
Kimi K3 fuels US AI anxiety
A new Chinese model is approaching OpenAI's capabilities.
Anthropic settlement wins final approval
The court rejected objections to the payout amount.
Adobe Firefly AI video editor review
An excellent choice for AI image creation, backed by powerful editing features and a generous free plan.
π°
Huawei launches Pura 90s with AI imaging
The new flagship series features 5.5G connectivity and AI-powered imaging capabilities.
Adobe Firefly AI video editor review
An excellent choice for AI image creation, backed by powerful editing features and a generous free plan.
Chinese AI agent beats Claude Code
Zhejiang Universityβs Qiushi Engine leads the ResearchClawBench, though reliability remains a challenge.
Anthropic settles copyright lawsuit
The $1.5 billion settlement impacts the future of AI training.
Adobe Firefly AI video editor reviewed
Firefly is an excellent choice for AI image creation, backed by powerful editing features.
π°
Huawei launches Pura 90s with AI imaging
The new series features 5.5G connectivity and AI-powered camera features.
Google's 'Frozen v2' chip for Gemini
The new server chip aims to improve Gemini model speed and efficiency.
US judge approves Anthropic's $1.5 billion settlement
The court approved the agreement, resolving claims of book piracy.
Google expands Gemini 3.5 line
Google rolled out Gemini 3.6 Flash and two new 3.5 Flash models.
Anthropic settles copyright lawsuit
Anthropic has settled a class action lawsuit for $1.5 billion over AI training data.
Anthropic settles copyright lawsuit for $1.5B
The landmark ruling sets a precedent for AI training data.
Anthropic copyright case settlement details emerge
The US Court issued a settlement in the Anthropic copyright case, impacting AI model training.
Will Kimi K3 create another DeepSeek moment?
Kimi K3 is a new Chinese AI model attracting attention.
Humanoid raises $152M for humanoid robotics
The London-based company is now Europeβs first pure-play robotics unicorn.
Googleβs βFrozen v2β AI chip
Alphabet is reportedly developing a next-generation chip optimized for Gemini models.
Google develops custom AI chip Opal
The chip aims to boost Gemini efficiency.
Google releases smaller Gemini AI models
Gemini 3.6 Flash and 3.5 Flash-Lite boost AI agent efficiency amid 3.5 Pro delays.
Google develops custom AI chip for Gemini
Alphabet is building a specialized chip to improve Geminiβs efficiency.
π°
Google releases three new AI models
The release includes a powerful model and one fine-tuned for cybersecurity.
Anthropic copyright settlement final
A federal court approved the $1.5 billion settlement with authors and publishers.
China opens AI models to catch up
Chinese companies are taking a different approach to AI development than the US.
Google advances AI push with new chip
Google plans a next-generation chip to improve Gemini performance and lower costs.
Claudeβs language differences
A study found Claude uses more polite phrasing in Arabic than in other languages.
π°
Agentic AI will change operating models
Experts discuss how agentic AI will impact organizations.
US Considers Ban on Chinese AI Models
The US government is weighing a ban on Chinese AI models, reflecting concerns about technological dominance.
Googleβs Opal AI Builds Apps From Your Words
Googleβs Opal AI can build applications from natural language prompts, though its reliability is still variable.
Anthropic settles copyright lawsuit
A US judge approved Anthropic's $1.5 billion settlement.
Does China's latest AI model equal US rivals?
Scientists assess the capabilities of Kimi K3.
NVIDIA unveils DLSS 5 and Cosmos AI
New technologies were showcased at SIGGRAPH.
HP HyperX OMEN 15 review
A gaming laptop with a powerful processor and display.
Cisco security AI competes with Anthropic
Ciscoβs open-weight models claim to outperform OpenAI and Anthropic in cybersecurity.
Google AI recreates Pele goal
Google used live stadium capture and archival material to bring a legendary, never-filmed goal to life.
Anthropic Settlement Wins Final Approval
The $1.5 billion copyright deal is now official.
Claude Code rewrites Bun in Rust
Anthropic's Claude Code successfully rewrote the Bun runtime from Zig to Rust in just 11 days.
ChatGPT loses market share
ChatGPT's audience share dropped below 50% as Google Gemini and Claude gained traction.
π°
Kimi K3 could boost chip demand
The model may drive enterprise workloads for Nvidia and Micron.
Anthropic settles copyright lawsuit for $1.5B
A landmark agreement addresses concerns about AI training data.
Anthropic's book-piracy settlement wins approval
The deal is believed to be the largest copyright settlement in US history.
Google develops new AI chip for Gemini
Alphabet is working on custom hardware to improve the efficiency of its Gemini models.
Infinity raises $15M for AI inference software
Infinity Inc. secured seed funding to develop software enabling AI inference on diverse chipsets.
π°
Google's 'Frozen v2' chip arrives in 2028
Google is designing Frozen v2, a chip expected to be 6-10 times more efficient, enhancing Gemini by 2028.
Google stock jumps on AI chip news
The company is developing a chip to optimize Gemini.
Alibaba intensifies AI coding competition
The company is challenging Anthropic's dominance in the coding space.
Meta shifts towards becoming a compute provider
Zuckerberg is quietly investing in infrastructure for AI.
Googleβs Frozen v2 chip integrates Gemini
Google is developing a new server chip to embed Gemini AI directly into the hardware.
π°
Anthropic receives court approval
A US federal judge granted final approval to Anthropic's $1.5 billion settlement.
Anthropic settles copyright lawsuit for $1.5 billion
A US judge approved Anthropicβs settlement in a copyright case over AI training data.
OpenAI paused ErdΕs model after sandbox escapes
Security concerns led to the suspension of the unreleased model.
Chinese AI model rivals ChatGPT
A new Chinese AI model is surprising Silicon Valley, indicating rapid progress in the region.
Chinese AI model could trigger sharemarket collapse
Fund managers warn MoonShotβs Kimi K3 could cause a market downturn.
Hugging Face used a Chinese AI model to defend against a cyberattack
US model guardrails hindered their defense, forcing them to use a Chinese alternative.
Hugging Face used a Chinese AI model for cyber defense
US model guardrails hindered its ability to respond to an autonomous attack.
Anthropic settles copyright lawsuit for $1.5B
A federal judge approved the landmark settlement.
π°
Alphabet stock jumps on AI chip news
Alphabet shares rose after reports of Google developing a new AI chip for Gemini.
π°
Infinity's Ignition agent automates AI chip software
The agent rapidly engineers custom inference layers for new silicon.
EQX Flow generates viral videos
The platform uses AI to create videos for communication purposes.
Googleβs βFrozenβ chip boosts Gemini
Alphabet stock rose on reports of a new chip embedding Gemini architecture.
Kimi K3 exposes GPU wall
Moonshot AIβs Kimi K3 highlights the importance of Nvidia and Microsoft.
AI re-writes enterprise storage security rules
AI is increasing pressure on enterprise data, demanding new security approaches.
Netflix used AI in 300 releases this year
Generative AI workflows were primarily used in post-production.
Hugging Face uses GLM-5.2 for breach analysis
Google is reportedly developing a new AI chip for Gemini.
π°
Kimi K3 could face a US ban
Washington is considering restrictions on the Chinese AI model.
Google's Frozen V2 targets Gemini efficiency
The next-generation chip is designed to boost Gemini's performance and ease compute strain.
Hugging Face defends with open-weights
Hugging Face used Z.ai GLM 5.2 to defend against an attacker after a commercial model refusal.
NVIDIA AI detects deepfakes in milliseconds
NVIDIAβs new Synthetic Video Detector achieves up to 92% accuracy in identifying AI-generated videos.
Googleβs βFrozen v2β chip
Google's new server chip could arrive as early as 2028 and deliver six to ten times better performance.
Google shares rise on chip news
Alphabet stock gained on reports of a new server chip for Gemini.
π°
Infinity secures $15M seed funding for AI chip inference software
The startup focuses on enabling AI workloads across diverse chipsets.
π°
Alibabaβs Qwen 3.8 Max challenges Kimi K3
The AI and technology industry is seeing fast-moving developments.
Google Vids edits videos with prompts
Google's Vids now allows video editing through text prompts and custom avatar creation using Gemini Omni.
π°
China vs. US AI: Kimi K3 assessment
Moonshotβs Kimi K3 may not be a game changer yet.
Alphabet stock rises on chip news
The new AI chip, "Frozen v2," embeds Gemini architecture directly into silicon.
Google's Gemini chip integrates AI directly
A new server chip, nicknamed Frozen v2, could arrive in 2028 and boost Gemini performance.
Inkling AI model specs released
Thinking Machines' Inkling, a 975B parameter open-weights model, is now available; benchmarks and pricing are detailed.
OpenAI downplays token costs of open-weight models
Chair Bret Taylor argues frontier models are more efficient.
Moonshot AI's Kimi K3 launches with 2.8T parameters
The massive model rivals Claude and GPT in performance.
π°
EU Parliament builds its own AI platform
The European Parliament is creating an in-house AI platform with access to models from major providers.
Apple sues OpenAI
The lawsuit comes ahead of ChatGPT 6 release, intensifying competition.
π°
Decartβs Lucy 2.5 Advances Live Video Effects
Improved physics engine strengthens world model simulations.
Kimi K3 pauses registrations due to high demand
Overwhelming interest is straining Moonshot AI's capacity.
ByteDance Seedance 2.5 creates 30-second AI videos
Seedance 2.5 can generate longer AI videos than most commercial models, potentially useful for marketing content.
Kapwing AI video editor reviewed
Kapwing is a comprehensive AI video editor with both pros and cons compared to competitors.
Googleβs βFrozen v2β chip for Gemini
Google is reportedly working on a new server chip optimized for its Gemini models.
Rock enthusiast builds Fable threat
Yang Zhilin, a rock fan who rejected Apple, built Moonshot AI into a low-cost challenger to OpenAI and Anthropic.
White House to control frontier AI access
A new program would give the US government control over access to the most advanced AI models.
Kapwing AI video editor reviewed
A thorough review of Kapwingβs pros and cons compared to other AI video editing platforms.
π°
Traders predict new Claude Opus model release
A prediction market suggests an imminent launch.
Nvidia launches its own AI model
This could impact competition in the AI industry.
π°
Concerns about Chinese models are rising
Experts debate the impact of Chinese AI models on the frontier labs.
InVideo AI is a powerful video editor
InVideo AI offers a conversational AI assistant alongside its video editing tools.
Claude Fable 5 billing splits
Max and Team Premium subscribers retain access to the model.
Veed AI excels in subtitles and avatars
Veed AI is a polished platform with realistic avatars and 4K export capabilities.
InVideo AI video editor (2026) review
A powerful AI video platform with one of the most capable conversational AI assistants we've tested.
InVideo AI excels in conversational assistance
The platform offers a powerful AI video editor with a capable AI assistant for content creation.
Veed AI video editor reviewed
Veed is a polished AI video platform with strong subtitle and avatar features, but its free plan is limited.
New Chinese AI Rivals OpenAI & Claude
Moonshotβs Kimi K3 is approaching the performance of leading models.
π°
Kimi K3 drives China's AI advancements
High demand for Moonshot AI's model highlights China's rapid progress in artificial intelligence.
Alibaba launches Qwen 3.8 LLM
Alibaba's Qwen 3.8 model rivals Anthropic's Claude Fable 5, signaling a push in AI coding capabilities.
π°
Kimi K3 narrows gap with US rivals
Moonshot AIβs Kimi K3 is performing on par with Claude Fable 5 and GPT-5.6.
Gemini 3.5 Pro release delayed
Bloomberg reports coding issues are pushing back the release of Googleβs Gemini 3.5 Pro.
π°
China narrows AI gap with Kimi K3
The new model is ranked on par with Claude Fable 5 and GPT-5.6 Sol.
Kimi K3 sparks anxiety in Silicon Valley
Beijing-based Moonshotβs Kimi K3 model is narrowing the gap with US AI capabilities.
European Parliament adopts more AI
The institution is opening access to commercial models to aid lawmakers.
Google's Frozen v2 chip could be 10x more efficient
The custom AI processor aims to significantly improve Gemini's performance and address growing compute demands.
AI agent smartphones emerge in China
Tech firms are developing AI-powered smartphones with revolutionary mobile capabilities.
EU Parliament adopts AI for lawmakers
The institution is opening access to models from OpenAI, Meta, Anthropic and Mistral.
Anthropic revamps Claude subscriptions
Access to Fable 5 is changing for users.
Anthropic settles $2B copyright case
The settlement valuation hits $1.25T by December.
Google develops custom AI chip
Google is building a chip to boost Gemini efficiency.
Bundesdruckerei's MΓΆve project tames authority AI
Bundesdruckerei aims to ensure reliability and prevent faulty decisions in digital offices with its AI 'TΓV'.