Goldman Sachs funds AI infrastructure
The bank is aggressively bankrolling the buildout of data centers and AI hardware.
ddNews Β· a daily AI briefing for Dirk (generative video/ddStudio, Lit
The bank is aggressively bankrolling the buildout of data centers and AI hardware.
2,056 robots from 16 countries compete under stricter autonomy rules.
Users can now generate interactive Kanban boards and dashboards directly from spreadsheet data.
The upcoming public offering will be a major test of whether AI growth justifies extreme valuations.
The "Ultrafast" preview delivers 14x the speed of standard models using wafer-scale chips.

New demonstrations show the humanoid navigating concrete stairs and outdoor environments.
Both firms are cutting prices as cheaper Chinese rivals gain market share.

Experts warn that agent-led operations against infrastructure change the nature of cyberwarfare.
Hyperscalers are accelerating spend ahead of Nvidia's next-gen chip shipments.
New iterations of the video model show a significant creative leap in realism.
Hyperscalers accelerate spending in anticipation of Nvidia's next-gen architecture.
New chatbot allows users to transform sounds and generate custom plugins within a DAW-like environment.

The 27B parameter model under Apache-2.0 aims to beat larger models in programming and office tasks.

Agents using GPT-5.6 and Claude Opus 4.8 failed to independently produce high-quality NeurIPS papers.

A new inference mode delivering 750 tokens/sec via Cerebras hardware to court enterprise users.
A new workhorse model optimized for coding and agent workflows, released just three weeks after 3.6 [70, 76, 98].

Security researchers warn that stolen credentials from a March breach remain active.
A 30B parameter open-weight model designed to run entirely on local hardware.
The move aims to integrate optimization technology to lower the cost of running Claude [50, 118].

Strategic collaboration focuses on AI factories, mobility, and humanoid development.

Ukraine claims the S-71 'Monochrome' uses Orin NX modules for terminal guidance.

Gemini users can now toggle off visible watermarks for AI images, videos, and music.

Analysis shows agent projects fail when systems cannot reliably execute what the model knows to do.

OpenAI and Anthropic are cutting prices to fend off cheaper, high-performing Chinese rivals.

Claude Code is being used internally to manage app maintenance and dead-code removal via Slack.
A new open-weights model from China that claims to outperform GPT-5.6 Sol in coding benchmarks.
New model optimized for coding and agentic workflows, released just three weeks after its predecessor.
A new project uses text-native web fonts to block AI scrapers while remaining human-readable.
The new flagship model targets the US-China competition in open-source infrastructure and pricing.
New hardware claims 98% power reduction compared to Nvidia for 2027 production.
Google's latest low-cost model targets production-ready coding and autonomous workflows.

Anthropic implements machine-readable marks on all text and code to comply with the EU AI Act.
AI-powered humanoid robots are being deployed to assist teachers in South Africa.

A new open-weights model from China claiming 50% improvement in coding.
Tau Robotics is now deploying humanoid cleaners at $30/hour.
Anthropic research shows AI agents can clash or collude when assigned the same task.
Google's new model targets coding and agent workflows with a rapid release cycle.

A rare cross-border collaboration to train a custom AI model specifically for the Chinese market.
Rapid revenue growth for Claude has investors projecting a valuation that could eclipse SpaceX.

Jensen Huang proposes a new financial model to maintain GPU value and fund buildouts.
Both firms are slashing prices as Chinese rivals gain ground.

The new bipedal robot is built on NVIDIA's Isaac GR00T reasoning platform.

A new bipedal robot built on NVIDIA Isaac GR00T will be unveiled in Q1 2027.
OpenAI slashed GPT-5.6 Luna prices by 80% as Chinese rivals gain ground.
The company raised $5 billion to fuel its expensive AI data infrastructure push.
The Claude maker is reportedly exploring a $6 billion acquisition of the Nvidia-backed startup.
Anthropic research shows multi-agent systems can clash and collude without user knowledge.
A new workhorse model designed to outperform Claude Sonnet 5 and GPT-5.6 in coding and agentic workflows.

OpenAI launched a new mode increasing GPT-5.6 Sol's speed by 14x.

New research shows firms are deploying autonomous agents despite gaps in risk control.

A new system based on GLM-5.2 provides deployment-ready capabilities at lower prices.
The new ultrafast tier outperforms traditional GPU clouds in inference speed.
Huawei is on track to eclipse Nvidia within the Chinese domestic market by 2026.

A new service tier running up to 14x faster, powered by Cerebras hardware.

An autonomous agent that optimizes inference workloads for any AI chip in a matter of days.
A new workhorse model optimized for coding and autonomous agent workflows.
A new workhorse model optimized for coding and agent workflows.
A new workhorse model optimized for coding and agent workflows, claiming to beat Claude Sonnet 5.

The platform adds MIDI import, 32-bit export, and a chat-based interface for instrument creation.

IBM will certify thousands of consultants on OpenAI tech to accelerate enterprise adoption.

New multimodal inputs and native audio are now available for longer AI video generation.
Advancements in generative media are positioning Chinese firms to lead the "world model" boom.

Competitive edge in video generation is paving the way for advanced "world models."
Anthropic implements text and image watermarks to comply with the EU AI Act.
Anthropic is adding machine-readable marks to all text and image outputs to comply with the EU AI Act.
Anthropic is applying machine-readable marks to all text and image outputs.
The investor argues that customers are carrying too much risk for the hardware buildouts.
All Claude text and image outputs will now carry machine-readable marks for EU AI Act compliance.

A new high-performance model for coding and agents that reportedly outperforms Claude Sonnet 5 and GPT-5.6 Terra.

Research shows AI agents can clash or collude when assigned the same task, complicating safety tests.
Funding will scale products to handle security for AI models and applications.
Users can now inspect the "For You" engine to check for shadowbanning.
After a 1,000-hotel trial in India, the platform deploys PMS-native agents for independent hotels.

New research pushes toward live-streamed human simulation without iterative rendering.
A reported $6 billion deal aims to reduce the computing costs of running Claude models.
A potential $6B acquisition aimed at reducing the computing costs of running large models.
The new model and harness aim to reduce token spending by 52% for Fortune 500 users.

The "M.AX" initiative aims to transform regional auto industry production via AI.

The 5.2-foot robot will serve as a showroom assistant; Xpeng plans to scale its Iron robot to 1,000 units/month.
New concerns emerge regarding the legal proof of authorization when autonomous agents spend user money.

The Chinese lab releases its flagship model and open-sources its "Harness" agent software.

A strategy to ensure aging GPUs maintain value by incentivizing new AI buildout financing.

Jensen Huang warns that optimizing models for Huawei silicon over Nvidia chips would be a "horrible outcome."
A new tool focused on final-stage upscaling for AI-generated footage.
New startups are racing to catch up with China's mass production of humanoid robots.
Industry experts from Agility, Apptronik, and Persona AI are convening to discuss the future of humanoid hardware.
New tool specifically designed to upscale AI-generated footage for professional workflows.
Jensen Huang warns of a "horrible outcome" if China optimizes models specifically for local hardware.
User replaces Perplexity with a local model for instant offline search.
A major breach has leaked credentials for over 2,500 companies, including AWS and Samsung.

Positioned as the most intelligent open-weight model for its specific size.

The platform is shifting toward a full DAW with MIDI import and 32-bit export.
The Claude maker is reportedly exploring a $6 billion acquisition of the Nvidia-backed startup.
The new infrastructure allows GPT-5.6 Sol to hit 750 tokens per second.

The new model rivals GPT-5.6 Sol and Claude Fable 5 in coding and agentic workflows.
New systems identify irregular items with a capacity of 1,200 parcels per hour.
The frontier lab signals its next major leaps as competition with xAI and Google intensifies.
In partnership with six banks, Nvidia is treating GPUs as a long-term investable asset class.
Taiwan reports that agents like "Open Claw" compromised government records in just four days.
153GB of cloud credentials from 2,500 organizations, including AWS and Samsung, were leaked.
Taiwan reports that agents like Open Claw helped hack government networks in four days.
Strategic focus on connecting GPUs and racks to scale AI systems further.

Despite being the most powerful, Anthropic's Fable 5 is seeing low enterprise adoption due to high costs.
A breach exposed 153GB of cloud credentials from 2,500+ companies, including AWS and Samsung.
Anthropic is implementing text and image watermarks to comply with the EU AI Act.
New hardware featuring Tensor G6 and deep Gemini AI integration.
Claude will now watermark all generated text and images to comply with the EU AI Act.
A new toolkit allows Gemma language and vision models to run locally on Raspberry Pi hardware.
An 80% price cut aims to make high-volume enterprise workloads more scalable.
Hinton, Fei-Fei Li, and Andrew Ng argue for open access to maintain US competitiveness against China.
A record-breaking oversubscribed IPO signals massive market appetite for humanoid hardware.

New benchmarks place Grok 4.6 on par with GPT-5.6 Sol and Claude Opus 5.

OpenAI-backed venture focuses on bringing frontier AI capabilities to the enterprise sector.