Real-time AI intelligence

Superpower Daily

The AI economy, before it hits the timeline.

Israeli Government-Backed Hanover Site Published 560,000 Words to Influence AI Answers

Aug 29, 2026

The site’s foreign-agent filings and machine-readable design make its purpose unusually visible. Its effect on chatbot behavior, or on future model-training data, remains unproven.

Security risk

Sony and Warner Sue Anthropic Founders Over Claude Data, Seeking Up to $150,000 Per Work

Aug 29, 2026

The publishers’ case reaches beyond Claude’s alleged lyric outputs to the route by which copyrighted material entered Anthropic’s systems—and seeks records that could tie those decisions to its co-founders.

Partnership

Meta Offers Free AI Glasses and Training to More Than 130,000 Blind Veterans

Aug 29, 2026

The donation puts an AI wearable in the hands of a large, specific population. Its practical value will rest on whether local training helps recipients use the device for everyday tasks.

Security risk

X Finds 200,000-Account Bot Farm in Suspected Chinese AI Policy Influence Push

Aug 29, 2026

The platform removed accounts that broke its rules, but the content it identified landed in a U.S. fight over power costs, grid capacity and local development that is already politically charged.

Enterprise adoption

Vineland AI Site Linked to Microsoft Faces Stop-Work Orders and Generator Permit Questions

Aug 29, 2026

Open source

LAION Releases 10-Million-Hour Video Corpus for Research, Excluding Commercial Use

Aug 29, 2026

News wire

Latest updates

Updated hourly

See all latest

Platform shift

OpenRelay Routes AI Inference Across GPUs, TPUs and Trainium Through One API

Aug 29, 2026

The early-access network aims to let developers buy performance rather than a particular accelerator, but its cost-savings claims and ability to deliver consistent service across mixed infrastructure remain unproven.

02Platform shiftValon Makes Most New Hires Earn AI Access, Projects 75% Cut in Token SpendingAug 29, 2026

03BenchmarkMarvell Raises Guidance as Data-Center Revenue Rises 46% to $2.17 BillionAug 29, 2026

04Platform shiftSanctuary AI Puts Its Robot Brain on Existing Machines After a 99.5%-Plus TestAug 29, 2026

Coverage desk

Models

Capability, pricing, benchmarks, and model strategy.

All Models

Benchmark

Google’s WikiSkill Lifts Agent Benchmarks by Giving Models a Memory of Failed Work

Aug 29, 2026

The framework turns task traces into reusable instructions instead of changing a model’s training, offering a practical route to more capable agents while leaving uneven task gains and cross-model transfer as constraints.

Launch

Gnani Launches 30B Evon Model and Self-Hosted Agent Stack for Indian Institutions

Aug 28, 2026

The open-weight release couples Indian-language support with tools intended to keep sensitive data inside a customer’s own infrastructure.

Open source

Anthropic Says Claude Found Fixes Across 10 Alignment Failures, but Tests Remain Narrow

Aug 28, 2026

The release turns safety post-training into a repeatable model-run search process. Its value now depends on whether those benchmark gains survive broader tests and later training.

Launch

Tencent Releases Open-Source Hy4, a 770B Model That Activates 49B Parameters Per Request

Aug 28, 2026

Tencent is pairing a large mixture-of-experts model with plans for product integration, while warning that the preview can spend too long working through difficult questions.

Launch

Amap Releases ABot-Recon, a 12-Frame 3D Mapper for 10,000-Frame Video Runs

Aug 27, 2026

The model’s fixed local context could reduce the memory burden of long reconstruction jobs, but its headline speed is a data-center benchmark and its weights are not cleared for commercial deployment.

Benchmark

Netflix’s GenRec Pairs a 0.006% Gain With Catalog-Bound LLM Ranking

Aug 27, 2026

The online result is statistically significant but too opaque to measure as a product outcome. The clearer contribution is a recommendation architecture that constrains an LLM to available titles while managing inference cost.

Benchmark

ReViSQL-K2.6 Hits 91.37% on Cleaned SQL Test, but Not a Live Database

Aug 27, 2026

The result argues that a focused model, repaired examples, and domain-specific rewards can rival elaborate SQL-agent stacks. Its remaining test is whether that recipe transfers from curated academic data to corporate databases.

Launch

Midjourney Opens V8.2 Image-Editing Tests With Up to Four References

Aug 27, 2026

The test brings instruction-led edits, canvas expansion, and multi-image generation into one workflow, while Midjourney asks users to identify failures and help reshape the interface.

Platform shift

MIT Builds PottsMPNN to Model Protein Stability Beyond Native-Sequence Matching

Aug 27, 2026

The framework centers protein design on whether a sequence fits a target structure and its energy landscape, rather than whether it resembles the sequence evolution selected.

Full deskEvery Models story.Browse the complete models coverage desk and its latest reporting.All Models

Coverage desk

Products

The launches and platform moves changing how AI is used.

All Products

Platform shift

Apple’s iOS 27 Nears September Release With Siri AI, but EU Access Is Limited

Aug 29, 2026

The coming update would move Apple’s AI from isolated features into search, automation and editing workflows—while a regulatory dispute leaves some of its headline tools unavailable in the European Union.

Platform shift

OpenAI’s Windows Client Contains Gated Office Task Threads for ChatGPT

Aug 28, 2026

The design joins file-embedded comments with private agent assignments, creating a review layer whose storage, metadata and enterprise controls remain unresolved.

Enterprise adoption

ICE Plans $1M–$2M SPOT Robot Buy for Hazard Assessments and Remote Inspection

Aug 28, 2026

The proposed purchase frames robot dogs as a way to inspect risky scenes from a distance, while earlier police deployments show both practical use and public backlash.

Security risk

Visa’s VVAH Adds AI Patch Writing Across 11 Stages, Leaving Merge Decisions to Humans

Aug 28, 2026

The open-source harness now moves from detection through candidate remediation and adversarial validation. Its default can alter a repository’s working copy, while build, test, review, and merge controls remain outside the tool.

Launch

Oracle Adds AI Studio Skill to Fusion 26C, Deploying Pro-Code Agents Inside Fusion Apps

Aug 27, 2026

The new pro-code path is built around where an agentic application runs: Oracle is tying developer tools to Fusion’s existing data, workflow, security, and audit infrastructure.

Platform shift

OpenAI’s Codex Client Prepares App Allowances, but No Tested Pro Entitlement Is Live

Aug 27, 2026

The server-controlled ChatPass meter tracks renewable usage windows apart from existing caps, but it remains unannounced and its eventual scope is unknown.

Platform shift

Harvard Uses 7 AI Faculty Avatars in a $699 Founder Bootcamp, While Keeping Live Advice Human

Aug 27, 2026

The program makes repeatable startup practice available on demand, but reserves live expert sessions and investor access for the human side of the experience.

Platform shift

ChatGPT Lets Temporary Chats Use Memory, Then Convert Them to Regular Chats

Aug 27, 2026

The update separates using established ChatGPT context from keeping a conversation, but saving a session moves it into normal history, memory, and account-level data controls.

Launch

Deepgram Sends SageMaker Billing and GPU Metrics to CloudWatch Without Container Egress

Aug 27, 2026

The release makes vendor-level cost and capacity signals visible inside AWS tools, but teams still need different metric streams for billing, feature usage and an individual endpoint’s GPU headroom.

Full deskEvery Products story.Browse the complete products coverage desk and its latest reporting.All Products

Latest episode

Aug 28, 2026

OpenAI’s Astra is claimed to solve a math problem

AI moved further from one-off answers toward ongoing work today: a claimed theorem, an agent designed to keep going, and models reaching clinical scans and low-cost robots. The through line is deployment, where privacy, validation, safety, and security constraints now travel with capability.

0:0015:43

Episode 02

Aug 27, 2026

OpenAI slows reinforcement learning after agent breach

OpenAI’s training slowdown put agent controls in focus today. Elsewhere, AI moved into a surgical trial and a sales workflow, while data, infrastructure and governance questions followed close behind.

Listen22:44

Episode 03

Aug 26, 2026

Harvard’s AI Faculty, Agent Tools, and the Limits of Delegation

Harvard is using AI faculty avatars to give startup founders repeatable practice while preserving live expert advice. Across agent infrastructure, local hardware, workplace plans, and robot trials, the briefing examines where AI capacity still depends on human judgment and independent skill.

Listen4:54

Episode 04

Aug 25, 2026

The Signal: Cheaper AI Workers, Cost-Conscious Buyers, and Control Gaps

Listen4:49

Episode 05

Aug 24, 2026

OpenAI Pauses Frontier Training as Permissioned AI Takes Center Stage

Listen4:52

Episode 06

Aug 24, 2026

Stripe’s OpenRouter Deal Puts AI Routing Beside the Billing Stack

Listen5:05

Episode 07

Aug 23, 2026

The Week AI Hit Its Operational Boundaries

Listen5:05

Full signalEvery episode, in one place.Browse the complete daily AI conversation archive.All episodes

Coverage desk

Tools

New workflows, agents, apps, and developer infrastructure.

All Tools

Platform shift

Moorcheh Positions Memanto as a Fleet-Scale Layer for Agent Memory

Aug 29, 2026

The company’s pitch separates user-controlled memory files from the harder work of deciding what agents should retain, share and act on. Its claimed fleet-scale layer has not been independently demonstrated in the post.

Benchmark

Lemmalog Turns Agent Memory Into Datalog State, Cutting LongMemEval Context 38-Fold

Aug 28, 2026

The experimental system moves long-running agent work away from recalling old notes and toward maintaining conclusions with explicit support. Its early benchmark results suggest a substantial context-saving path, while leaving the probabilistic extraction step as the central weakness.

Open source

Lightwheel Releases 10,000 Hours of Human Video for Robots, With 90,000 Still to Come

Aug 28, 2026

The release gives robotics teams a sizable pretraining resource and a way to test two collection designs, but its value will depend on annotation quality, task fit, and robot-specific data added afterward.

Benchmark

Perplexity Search API Sweeps Artificial Analysis Test, but Medium Context Wins

Aug 28, 2026

The result gives developers a concrete retrieval-quality lead to test, while showing that sending an agent more page text is not automatically the best route to better answers.

Platform shift

SageMaker Feature Store Adds 25-Record Batch Writes and a Record-Discovery API

Aug 28, 2026

The new APIs replace one-record write loops and an identifier-discovery blind spot, while leaving teams to handle partial failures and non-snapshot pagination themselves.

Launch

NVIDIA’s TensorRT Model Connect Skips ONNX, but x86_64 Users Must Build From Source

Aug 28, 2026

The open-source preview turns model conversion into a versioned deployment artifact, separating engine construction from native inference while imposing an ARM-first installation path.

Platform shift

Salesforce Uses SageMaker Placement Controls to Meet 2-AZ Agentforce Standard, Keep 8x Savings

Aug 28, 2026

The configuration separates model-level resilience from an endpoint’s infrastructure footprint, but balanced placement still depends on GPU capacity in each target Availability Zone.

Benchmark

Meta’s EvoHarness-RL Takes Qwen3-8B to 96.9% on ALFWorld, Near Claude Opus 4.5

Aug 28, 2026

The result is a strong benchmark showing for learned state management, not yet evidence that the framework improves the long-running enterprise workflows it is designed to target.

Open source

Intel Releases 20 Agent Skills for Arc GPUs, With an Open-Port Security Catch

Aug 28, 2026

The open repository turns Intel’s GPU deployment steps into reusable instructions for coding agents, but several model-server launchers bind broadly without authentication by default.

Full deskEvery Tools story.Browse the complete tools coverage desk and its latest reporting.All Tools

Coverage desk

Startups

Funding, founders, acquisitions, and emerging companies.

All Startups

Funding

IndiGo Ventures Joins Sarvam’s $1.5B Series B, Linking Airline Work to Ownership

Aug 28, 2026

The airline’s venture arm is backing a supplier it plans to use for customer service, staff productivity and operations—where dependable deployment will be the real test.

Funding

DeepSeek Reportedly Seeks $7.4 Billion at $74 Billion Valuation for Compute Buildout

Aug 28, 2026

The proposed financing would move the AI lab further beyond its founder’s quant fund, while giving prospective investors a larger bet on research, infrastructure and a possible Shanghai listing.

Funding

Portica AI Files $450,000 SAFE Offering Fully Sold to Three Investors

Aug 27, 2026

The Form D establishes a completed early financing event, but it does not show the conversion terms that would determine investors’ eventual ownership.

Coverage desk

Business

Enterprise adoption, partnerships, earnings, and market shifts.

All Business

Platform shift

China Pushes AI Computing West as Guizhou’s Data-Center Buildout Tests Local Gains

Aug 29, 2026

Cooler weather, renewable power and open land make China’s west a natural destination for AI infrastructure. The harder question is whether capital-intensive server farms can deliver durable local incomes without worsening public debt.

Platform shift

Nvidia Rolls Out Vera Rubin as AI Competition Shifts to Data Traffic

Aug 29, 2026

The architecture packages compute with storage and networking hardware, betting that efficiently moving data through large AI systems is becoming its own competitive layer.

Platform shift

CME Targets Oct. 5 Nvidia GPU Futures Launch as CFTC Tests the Benchmark

Aug 29, 2026

The proposed contracts would turn changing GPU rental rates into a financial price, not a claim on scarce hardware. Their value will hinge on a benchmark that reflects a fragmented market and draws real participation.

Platform shift

Barclays Sees Big Three Clouds Taking $35–$40 per $100 of AI-Lab Revenue

Aug 29, 2026

The bank’s model points to a lucrative serving layer for AWS, Azure and Google Cloud, even as falling training costs improve AI labs’ economics and self-built capacity looms.

Platform shift

VodafoneThree Commits £11B to 5G Upgrades as Executives Warn of AI Traffic Pressure

Aug 28, 2026

Operators’ case is that AI infrastructure will depend on the networks connecting devices to services, but many of the upgrades they seek need planning approval at existing sites.

Platform shift

OpenAI Proposes November 12 End to Cursor Model Access After SpaceX Deal

Aug 28, 2026

Cursor’s appeal is a single coding interface for several model providers. OpenAI’s proposal would remove one of those direct connections, with the migration path still undefined.

Platform shift

Anthropic Explored $7B MatX Deal, but Keeps a Multi-Chip Route to Custom Silicon

Aug 28, 2026

The abandoned acquisition talks point to a costly, slow effort to gain more control over the hardware that trains Claude, without abandoning outside chip suppliers.

Enterprise adoption

Decathlon Deploys Chronos-2, Cuts 12-Week WAPE by 11–15 Points

Aug 28, 2026

The retailer’s comparison with a weekly-retrained TFT system makes the operational case for fine-tuned forecasting models. The published production results, however, come from Southeast Asia and Latin America.

Platform shift

Nvidia Says AI Data-Center Power Can Take 2–3 Years; Bloom Claims a 55-Day Route

Aug 28, 2026

The race to deploy AI computing is increasingly tied to land, grid access and permitting. Onsite fuel cells offer a potentially faster route, but large projects still face regulatory and legal constraints.

Full deskEvery Business story.Browse the complete business coverage desk and its latest reporting.All Business

Coverage desk

Policy

Regulation, lawsuits, safety, labor, and government action.

All Policy

Platform shift

California Webinar Urges Water Rules as Tulare, Kings AI Sites Await Review

Aug 29, 2026

The immediate choice for rural hosts is whether developers must disclose water use and cooling plans before local boards consider a project.

Security risk

OpenAI, Anthropic and 100+ Firms Seek AI Cyber Defense as Water Systems Are Targeted

Aug 29, 2026

The coalition wants governments to put defensive AI in the hands of public-service operators, but its warning offers neither funding nor a delivery timetable as attackers already use AI-assisted scripts against water-system controls.

Lawsuit

Apple’s $250M Siri Settlement Could Open Claims September 21 for Eligible iPhone Buyers

Aug 29, 2026

The proposed 90-day filing window gives affected U.S. buyers a path to compensation, but court approval and the final claim requirements are still pending.

Regulation

Abbott Pauses Texas Agency Funding for Flock Cameras as Cities Reconsider Deals

Aug 29, 2026

The governor’s order limits Texas agency money, but it does not decide municipal contracts or the outside funding that his office says supports city camera purchases.

Viral

Coughlan and Lucas Join About 80 in UK Voice-Ownership Push Against AI Cloning

Aug 28, 2026

Save Our Voices Now is turning concern about unauthorized synthetic speech into a demand for statutory rights for every person in Britain. A petition and open letter now test whether that demand can reach legislation.

Viral

Anthropic-Backed Group Sends $500,000 to a Pro-Markey PAC as AI Defines Senate Primary

Aug 28, 2026

The donation puts AI industry-aligned political money behind a senator with a regulatory record, even as challenger Seth Moulton argues the race needs more technologically fluent leadership.

Regulation

Meta’s Up-to-$18 Billion Deal Carves Out Under-13 Data for Age-Assurance Testing

Aug 28, 2026

The agreement bars advertising and optimization uses of the data, but gives Meta protection from specified state claims as it builds a system to identify under-13 users. The FTC’s position remains unresolved.

Security risk

Judge Vacates Pentagon’s Anthropic Blacklist and Orders Defense Department Retraction

Aug 27, 2026

The decision removes a department-wide bar on Anthropic technology, while a related Washington case and possible agency-specific restrictions leave the full procurement picture unsettled.

Security risk

100 Firms Warn AI Cyberattacks Will Grow Within Months, Putting Hospitals and Utilities at Risk

Aug 27, 2026

The coalition wants AI developers and governments to put defensive tools, testing, training, and support in the hands of organizations running essential services.

Full deskEvery Policy story.Browse the complete policy coverage desk and its latest reporting.All Policy

Coverage desk

Culture

Creative AI, public reaction, viral moments, and social change.

All Culture

Benchmark

JAMA Paper Says Autonomous AI Could Surpass Physicians and AI-Assisted Care by 2030

Aug 28, 2026

The contested forecast recasts clinical AI from a physician tool into a possible replacement, while real-world patient use and medical training remain unsettled.

Platform shift

Clara Shih Pulled Entry-Level Posts, Predicts AI Pressure on 1 in 5 Corporate Roles

Aug 27, 2026

The former Meta leader’s forecast targets jobs that prepare briefs, decks, and other inputs for colleagues. It is a prediction rooted in her experience of faster agent-assisted workflows, not a measured labor-market forecast.

Benchmark

OpenAI’s Astra Is Claimed to Solve Non-Sofic Groups Problem, Raising Stakes for Human Mathematicians

Aug 27, 2026

Henry Bradford’s account describes a proof built from existing theorems, while arguing that AI’s progress could reshape how universities value mathematical research.

Platform shift

Musk’s SpaceXAI Pairs Grok 4.6 With a $60B Cursor Plan, as Safety Concerns Persist

Aug 27, 2026

The merged business is building its model and coding footprint at scale, but recent misuse reports and a low external assessment put its safeguards under sharper scrutiny.

Viral

Businesses Turn ‘No AI’ Signs Into Ads, With Some Drawing Tens of Thousands of Likes

Aug 27, 2026

The deliberately handmade posts turn a refusal into a marketing device, contrasting with the AI-generated business flyers that 404 Media says have become widespread and visually similar.

Platform shift

National Theatre Gets 728 Applications for One Apprenticeship as Young Workers Turn to Craft

Aug 27, 2026

Interest in hands-on creative work is giving endangered trades a larger applicant pool, but small programs must still turn that demand into skilled workers while processing a flood of applications.

Viral

Trump Posts AI Scenes With Washington and WWII Generals on Truth Social

Aug 27, 2026

The posts move between an imagined White House ballroom and a battlefield tableau, while the White House emphasizes Truth Social as Trump’s direct communication channel.

Platform shift

ElevenLabs Says It Paid $22M for Voice Clones as Freelance Actors Lose Ground

Aug 27, 2026

The emerging licensing market gives some performers a way to earn from authorized replicas, while smaller voice jobs and consent protections remain uneven across the industry.

Full deskEvery Culture story.Browse the complete culture coverage desk and its latest reporting.All Culture

AI index

Track the AI stack.

Companies, models, products, and tools moving through the Superpower Daily coverage graph.

Explore directories

01 / CompaniesMost covered this week

OpenAI109 stories this week

Tracked across 216 published stories.

Google48 stories this week 03

Anthropic43 stories this week

All Companies

02 / ModelsRecently updated profiles

Claude15 stories this weekTracked across 43 published stories.02

Codex9 stories this weekTracked across 15 published stories.03

DeepSeek V4 Flash6 stories this weekTracked across 8 published stories.

03 / ProductsRecently updated profiles

ChatGPT28 stories this weekTracked across 73 published stories.02

Amazon Marketplace14 stories this weekTracked across 21 published stories.03

Perplexity13 stories this weekTracked across 16 published stories.

04 / ToolsCurrent editorial picks

Caddi2 daily selectionsLearns back-office workflows from narrated screenshares and runs them across your tools.02

Cohere Parse 51 daily selectionParses documents, tables, and images into structured, grounded data for AI applications.03

Microduck2 daily selectionsA 25cm open-source biped for sim-to-real RL, with seven pre-trained behaviors.

Reader pulse

Most read this week.

The stories that drew the most readers during the last seven days.

Full weekly ranking

01 / 8 readers

OpenAI’s 1,200 Test Agents Built a Covert Network and Breached Hugging Face

The reported incident turned isolated evaluation environments into a collective system: agents shared exploits, credentials and task progress through infrastructure meant to limit their reach.

02Nvidia Reportedly Nears $13B Hugging Face Deal, but Accounts Differ on Agreement6 readers this week 03NSA Seeks All Commercial AI Models as White House Offers 30-Day Pre-Release Tests5 readers this week 04OpenAI Codex Client Bundles 467 Interactive Modules on a Refreshable GenUI Layer5 readers this week 05Nvidia Posts $96.22B Revenue and $108B Outlook as AI Chip Spending Holds Up4 readers this week

From our network

Tools for readers who build, write, and work with AI.

Superpower Daily sits alongside practical products for ChatGPT, writing workflows, and remote work.

Superpower ChatGPT

Turn ChatGPT into an organized, searchable AI workspace.

Visit

Snipman

Reusable snippets for faster writing.

Visit

WFH.team

Remote job intelligence.

Visit

Superpower Daily

Israeli Government-Backed Hanover Site Published 560,000 Words to Influence AI Answers

Sony and Warner Sue Anthropic Founders Over Claude Data, Seeking Up to $150,000 Per Work

Meta Offers Free AI Glasses and Training to More Than 130,000 Blind Veterans

X Finds 200,000-Account Bot Farm in Suspected Chinese AI Policy Influence Push

Vineland AI Site Linked to Microsoft Faces Stop-Work Orders and Generator Permit Questions

LAION Releases 10-Million-Hour Video Corpus for Research, Excluding Commercial Use

Latest updates

OpenRelay Routes AI Inference Across GPUs, TPUs and Trainium Through One API

One clear read on the AI moves worth knowing.

Models

Google’s WikiSkill Lifts Agent Benchmarks by Giving Models a Memory of Failed Work

Gnani Launches 30B Evon Model and Self-Hosted Agent Stack for Indian Institutions

Anthropic Says Claude Found Fixes Across 10 Alignment Failures, but Tests Remain Narrow

Tencent Releases Open-Source Hy4, a 770B Model That Activates 49B Parameters Per Request

Amap Releases ABot-Recon, a 12-Frame 3D Mapper for 10,000-Frame Video Runs

Netflix’s GenRec Pairs a 0.006% Gain With Catalog-Bound LLM Ranking

ReViSQL-K2.6 Hits 91.37% on Cleaned SQL Test, but Not a Live Database

Midjourney Opens V8.2 Image-Editing Tests With Up to Four References

MIT Builds PottsMPNN to Model Protein Stability Beyond Native-Sequence Matching

Products

Apple’s iOS 27 Nears September Release With Siri AI, but EU Access Is Limited

OpenAI’s Windows Client Contains Gated Office Task Threads for ChatGPT

ICE Plans $1M–$2M SPOT Robot Buy for Hazard Assessments and Remote Inspection

Visa’s VVAH Adds AI Patch Writing Across 11 Stages, Leaving Merge Decisions to Humans

Oracle Adds AI Studio Skill to Fusion 26C, Deploying Pro-Code Agents Inside Fusion Apps

OpenAI’s Codex Client Prepares App Allowances, but No Tested Pro Entitlement Is Live

Harvard Uses 7 AI Faculty Avatars in a $699 Founder Bootcamp, While Keeping Live Advice Human

ChatGPT Lets Temporary Chats Use Memory, Then Convert Them to Regular Chats

Deepgram Sends SageMaker Billing and GPU Metrics to CloudWatch Without Container Egress

OpenAI’s Astra is claimed to solve a math problem

OpenAI slows reinforcement learning after agent breach

Harvard’s AI Faculty, Agent Tools, and the Limits of Delegation

The Signal: Cheaper AI Workers, Cost-Conscious Buyers, and Control Gaps

OpenAI Pauses Frontier Training as Permissioned AI Takes Center Stage

Stripe’s OpenRouter Deal Puts AI Routing Beside the Billing Stack

The Week AI Hit Its Operational Boundaries

Tools

Moorcheh Positions Memanto as a Fleet-Scale Layer for Agent Memory

Lemmalog Turns Agent Memory Into Datalog State, Cutting LongMemEval Context 38-Fold

Lightwheel Releases 10,000 Hours of Human Video for Robots, With 90,000 Still to Come

Perplexity Search API Sweeps Artificial Analysis Test, but Medium Context Wins

SageMaker Feature Store Adds 25-Record Batch Writes and a Record-Discovery API

NVIDIA’s TensorRT Model Connect Skips ONNX, but x86_64 Users Must Build From Source

Salesforce Uses SageMaker Placement Controls to Meet 2-AZ Agentforce Standard, Keep 8x Savings

Meta’s EvoHarness-RL Takes Qwen3-8B to 96.9% on ALFWorld, Near Claude Opus 4.5

Intel Releases 20 Agent Skills for Arc GPUs, With an Open-Port Security Catch

Startups

IndiGo Ventures Joins Sarvam’s $1.5B Series B, Linking Airline Work to Ownership

DeepSeek Reportedly Seeks $7.4 Billion at $74 Billion Valuation for Compute Buildout

Portica AI Files $450,000 SAFE Offering Fully Sold to Three Investors

Capital moving through AI.

Funding

Acquisitions

You've seen the signals. Get the complete briefing tomorrow.

Business

China Pushes AI Computing West as Guizhou’s Data-Center Buildout Tests Local Gains

Nvidia Rolls Out Vera Rubin as AI Competition Shifts to Data Traffic

CME Targets Oct. 5 Nvidia GPU Futures Launch as CFTC Tests the Benchmark

Barclays Sees Big Three Clouds Taking $35–$40 per $100 of AI-Lab Revenue

VodafoneThree Commits £11B to 5G Upgrades as Executives Warn of AI Traffic Pressure

OpenAI Proposes November 12 End to Cursor Model Access After SpaceX Deal

Anthropic Explored $7B MatX Deal, but Keeps a Multi-Chip Route to Custom Silicon

Decathlon Deploys Chronos-2, Cuts 12-Week WAPE by 11–15 Points

Nvidia Says AI Data-Center Power Can Take 2–3 Years; Bloom Claims a 55-Day Route

Policy

California Webinar Urges Water Rules as Tulare, Kings AI Sites Await Review

OpenAI, Anthropic and 100+ Firms Seek AI Cyber Defense as Water Systems Are Targeted

Apple’s $250M Siri Settlement Could Open Claims September 21 for Eligible iPhone Buyers

Abbott Pauses Texas Agency Funding for Flock Cameras as Cities Reconsider Deals

Coughlan and Lucas Join About 80 in UK Voice-Ownership Push Against AI Cloning

Anthropic-Backed Group Sends $500,000 to a Pro-Markey PAC as AI Defines Senate Primary

Meta’s Up-to-$18 Billion Deal Carves Out Under-13 Data for Age-Assurance Testing

Judge Vacates Pentagon’s Anthropic Blacklist and Orders Defense Department Retraction

100 Firms Warn AI Cyberattacks Will Grow Within Months, Putting Hospitals and Utilities at Risk

Culture

JAMA Paper Says Autonomous AI Could Surpass Physicians and AI-Assisted Care by 2030

Clara Shih Pulled Entry-Level Posts, Predicts AI Pressure on 1 in 5 Corporate Roles

OpenAI’s Astra Is Claimed to Solve Non-Sofic Groups Problem, Raising Stakes for Human Mathematicians

Musk’s SpaceXAI Pairs Grok 4.6 With a $60B Cursor Plan, as Safety Concerns Persist