AI news, Oct 22, 2026
AI news for Oct 22, 2026: 42 stories ranked by coverage, each with a summary and source links.
Top stories
- Products
1Password Launches Secure Login for AI Agents
1Password has introduced a new feature that enables AI agents to log into user accounts while keeping private credentials hidden from the agents. This aims to address security concerns associated with AI agents managing personal accounts, balancing automation needs with data protection.
The VergeTechCrunchiThome3 outlets
- Products
Uber Rolls Out ChatGPT Work to Employees
Uber announced the rollout of ChatGPT Work to more than 30,000 employees. The tool helps teams conduct research, analyze data, create presentations, and complete complex workflows to boost internal productivity.
- Policy
OpenAI Partners on Consistent Standards for AI Safety
OpenAI collaborates with distribution partners to enforce consistent safeguards and support responsible access. This aims to clarify expectations in usage policies, ensuring uniform safety standards across platforms.
- Products
Anthropic launches Claude Duet agent app
Anthropic launched the Claude Duet app, designed for AI agent collaboration. It is currently available via waitlist for Pro and Max plan users, signaling Anthropic's bet on agent-based products.
The VergeTechCrunch2 outlets
- Products
AI Shopping Agents Prompt Retailer Caution
Tech companies are promoting AI agents for shopping, marketing them as democratized personal assistants. Some retailers remain cautious, concerned that automated agent transactions may pose complex challenges to existing business models.
The Verge2 outlets
Models2
- Models
Recursive launches Flint coding model built by AI
Startup Recursive released Flint, a free, laptop-sized coding model. The model was largely built by the company's own AI research system, serving as an early test of AI self-improvement capabilities.
- Models
Ant Digital Launches Decision Model LingDT-Decision-1.0
Ant Digital released decision model LingDT-Decision-1.0 on October 22. The company claims the model has higher accuracy than competitors like Microsoft and Jev. This release expands the domestic AI product matrix in the decision-making sector.
Products11
- Products
AT&T Rethinks In-House Legal Work with OpenAI
AT&T is working with OpenAI to rethink in-house legal knowledge access. Its LegalEdge team is using AI to streamline intake processes and scale legal judgment to improve internal efficiency.
- Products
Check if Model Runs on AWS Trainium
Hugging Face has provided a guide or tool to help users check if their AI models are running on AWS Trainium hardware. This addresses the need for transparency in AI infrastructure and specialized chip integration.
Hugging Face2 outlets
- Products
Spotify bets on AI agents for music
Spotify co-CEO Gustav Söderström stated that the company's current AI strategy focuses on becoming the music service of choice for AI agents. He believes the rise of AI agents is positive news for Spotify, helping expand its licensing and distribution reach.
- Products
Eiliko Hires Security Firm After Vulnerabilities
The Eiliko AI chat charm was found to have security vulnerabilities, allowing hackers to access full conversation transcripts. Following media reports, the manufacturer announced it is hiring an independent security firm to assess and improve product security.
- Products
Meta says Muse AI app has over 10M users
Meta announced that its AI app, Muse, has surpassed 10 million users. The company noted that shopping and personal productivity are among the top activities on the platform.
- Products
Walmart allows AI agents to make purchases
Walmart has introduced a new technology that allows AI agents to make purchases on behalf of customers with spending controls. This contrasts with Amazon's blocking of AI agents, highlighting expanding AI applications in e-commerce.
- Products
NVIDIA Launches Minerva AIOps Platform
NVIDIA launched Minerva, an AIOps platform designed to enhance end-to-end observability in AI factories. The platform combines telemetry from infrastructure and AI workloads, applying purpose-built AI models to help teams detect issues early and maintain operational efficiency.
- Products
NVIDIA DSX Helps Break Power Bottlenecks
NVIDIA's DSX AI Factory Platform helps companies plan and simulate grids and AI factories as a unified system to address power bottlenecks in infrastructure. Over $300 billion in investment highlights the urgency of this issue, with the platform supporting operations from substations to racks.
- Products
Anthropic launches proactive AI agent Claude Duet
Anthropic released Claude Duet, a proactive AI agent for professionals and small businesses. It checks connected apps, organizes tasks, and sends reminders while offline. US Pro and Max subscribers can register for the research preview.
- Products
Salesforce's Taiwan Operations Post Double-Digit Growth
Salesforce's Taiwan General Manager stated that local revenue and new customers grew by double digits last year. Focus is shifting to agents and data infrastructure, with Agentforce comprising 40% of new deals, indicating AI applications moving from trials to business integration.
- Products
VS Code Supports Meta's Pyrefly Type Checker
Microsoft and Meta collaborated on a Type Server Protocol, allowing VS Code's Pylance extension to switch to Meta's Pyrefly for type checking. The feature is now available to all users, retaining autocomplete and Copilot integration.
Research7
- Research
Claude Failed to Invent Scorpion Antivenom
Investment firm Citrini Research partnered with Anthropic to test if an unreleased Claude model could create scorpion antivenom. The model produced promising starting points but failed to develop an actual drug, highlighting current limitations in AI-driven drug discovery.
- Research
IBM Granite Time-Series Ensemble Forecasting
Hugging Face released research on IBM Granite time-series models, detailing ensemble forecasting techniques to improve model performance. The publication offers technical insights into applying AI models for time-series data analysis.
- Research
NVIDIA Kumo Outperforms Supervised ML on Relational Data
NVIDIA released research stating that its Kumo Relational method outperforms supervised machine learning on relational data without requiring training. The study highlights potential advantages of the new architecture for specific data types.
- Research
OpenAI and Wiz discover critical PHP vulnerabilities
OpenAI and Google's Wiz team discovered long-standing security vulnerabilities in PHP. The incident raised online concern, demonstrating AI's role in uncovering security issues while highlighting potential risks associated with AI technologies.
- Research
SLM Vocabulary Specialization Cuts Inference Cost by 77%
A study indicates that specializing the vocabulary of Small Language Models (SLMs) can reduce inference costs by 77% in real-world applications. This offers a new optimization path for cost-effective AI development.
- Research
AI inference creates hidden data storage costs
MIT Technology Review reports that businesses often overlook the storage costs of backend data generated by AI inference. Billions of daily prompts and media outputs create persistent data needs, making storage an underestimated expense in AI operations.
- Research
AI Helps Discover Decade-Old PHP Security Flaw
Security firm Wiz utilized AI to discover PHP2Shell, a vulnerability existing for over a decade in PHP, which was patched by developers on October 21. This highlights AI's potential in uncovering legacy security issues.
Open source3
- Open source
NVIDIA Expands Cosmos for World Models
NVIDIA expanded the Cosmos platform with new training tools, simulation checkpoints, and data factory blueprints to help developers build world models for physical AI systems. The platform ranks in the top five on six open model leaderboards, strengthening its position in the physical AI sector.
- Open source
China Telecom releases open-source intent model Xing4.0-29B-Intent
China Telecom's Star AI Lab released and open-sourced the dedicated intent understanding model Xing4.0-29B-Intent. The model is designed to accurately identify user intent in complex scenarios, supporting multiple task formats. This expands the open-source AI ecosystem in vertical applications.
- Open source
Microsoft open-sources Scope for AI agent evaluation
Microsoft open-sourced Scope, a platform that evaluates AI coding agents like GitHub Copilot on defined tasks. It logs operations and results, helping teams verify correct use of SDKs and APIs and identify agent issues.
Funding9
- Funding
Nirva raises $33M for AI pendant device
Nirva has raised $33 million to develop its $249 ambient AI pendant, which is designed to listen only. The company expects to ship the device by the end of the year.
- Funding
DeepSeek reportedly raises $15B amid US restrictions
DeepSeek has reportedly raised $15 billion, marking one of the largest funding rounds in China. The news emerged days after the US ordered federal agencies to remove its technology over national security concerns.
- Funding
Blackstone exec says AI buildout debt fears are overblown
Jas Khaira, head of AI at Blackstone, stated that market fears over debt related to AI infrastructure buildout are overblown. He explained why the private equity giant is not panicked by market jitters. The comments reflect the firm's assessment of long-term financing risks in AI.
- Funding
Meta Plans New Stock Class to Fund AI Buildout
Meta is planning to create a new class of stock to help fund its broad investments in AI models and services. This move aims to provide a sustained source of capital for its large-scale AI infrastructure buildout.
- Funding
Meta issues new shares to fund AI push
Meta announced the issuance of non-voting shares to fund AI infrastructure expansion and potential acquisitions, with trading starting in 2027. The company cites the massive investment scale required to sustain growth and compete effectively.
- Funding
Blacksmith reaches unicorn valuation after $66M Series C
AI coding platform Blacksmith closed a $66 million Series C round led by Sequoia, valuing the company at $1 billion. The company helps enterprises build and test AI code more effectively, and the funding will support its growth.
- Funding
Tesla profit halved as Musk claims FSD dominance
Tesla's profit halved, causing its market value to drop below that of General Motors. Elon Musk stated that FSD will lead in the Chinese market, noting the feature gained 700,000 new paid users in a year.
- Funding
Meta issues Class C shares to fund AI investments
Meta announced the creation of non-voting Class C common shares in January 2027 to fund AI infrastructure, acquisitions, and employee compensation. The move maintains CEO Mark Zuckerberg's control and will result in a stock split for existing shareholders.
- Funding
Meta Q3 revenue up 26%; AI capex up to $145B
Meta reported Q3 2026 revenue of $64.6B, up 26% year-over-year. It forecasts annual capex of $135B–$145B to expand AI infrastructure, signaling continued heavy AI investment.
Policy2
- Policy
OpenAI and Nvidia record lobbying spend
Federal disclosures show OpenAI, Anthropic, and Nvidia recorded their highest lobbying spend in Q3. Nvidia spent the most at $2.2 million, followed by Anthropic at $2.1 million and OpenAI at $1.5 million, indicating increased policy engagement by AI leaders.
- Policy
Texas data centers may reverse air quality gains
Rapid construction of AI data centers in Texas is driving up power demand. Relying on gas for this electricity could revert air pollution to 2000s levels, undoing decades of air quality improvements and highlighting environmental costs.
Other3
- Other
OpenAI CLI Tool Draws Attention
OpenAI's Command Line Interface tool was discussed on Hacker News without detailed feature descriptions or release specifics, reflecting community interest in OpenAI's developer tools.
- Other
Supply chain analysis: Who supplies embodied data to OpenAI?
QbitAI published an analysis of the embodied intelligence (Physical AI) supply chain. The report explores which companies are entering the core global supply chain and providing training data to OpenAI and others. This highlights the importance of data resources in the embodied AI sector and integration trends.
- Other
LiteLLM hit by supply-chain attack; backdoor in latest release
Open-source AI gateway LiteLLM suffered a supply-chain attack: attackers stole an engineer's GitHub credentials to inject a backdoor and infostealer into v1.107.0. The maintainer has pulled the package and paused releases.
All news → · All rights belong to the original creators