News

October 23, 2026

Friday

  1. Crunchbase NewsStartups

    Tenstorrent and Flapping Airplanes lead the week’s largest U.S. funding rounds

    Tenstorrent raised $1.5 billion in Series E funding at a $6.2 billion valuation. Flapping Airplanes secured $1 billion in Series A funding less than 10 months after its $180 million seed financing.

    Why it mattersBuilders should expect well-funded competitors and potential infrastructure partners across AI processors, lower-cost models, robotics, biotech and manufacturing software.

  2. The VergeTech

    Google updates privacy policy to allow ad targeting from Gmail, Drive, and Photos linked to AI

    Google added an exception to its privacy policy regarding the use of Drive, Gmail, and Google Photos data for personalized ads, as spotted by Android Authority. Users who link Drive, Gmail, or Google Photos to AI Mode and other Search tools through Google’s Personal Intelligence feature may now receive ads targeted using that content. Users can stop this ad targeting by disconnecting specific apps from Search’s Personal Intelligence features or turning off Personalized Recommendations in the Search Services Personalization settings page.

    Why it mattersWhen building AI features that connect to user workspaces, separating context-sharing permissions from monetization and ad-targeting settings avoids forcing users to choose between assistant utility and data privacy.

  3. The VergeTech

    Spotify excludes 50 million AI slop and spam tracks from user recommendations

    Over the past year, Spotify excluded over 50 million AI-generated slop and spam tracks from its user recommendations. The platform detects suspected AI slop using listener engagement, user feedback, upload patterns, metadata, and third-party AI detection tools. Rather than requiring labels on all AI tracks, Spotify introduced an "AI Persona" badge for non-human artist identities and stopped recommending their music.

    Why it mattersBuilders running content marketplaces and recommendation feeds can protect discovery quality by combining behavioral signals, upload patterns, and metadata to demote low-effort AI content in ranking.

  4. The VergeTech

    Meta revises business terms on AI training, but legal experts say scraping risks remain

    Meta's new business account terms, effective October 30th, originally allowed using content from linked websites to develop, research, and train its products. Attorney Matt Topic notes the revised terms still let Meta use scraped content to "develop" any "business product," which is broad enough to include training AI. Meta spokesperson Daniel Roberts stated that site owners can still use robots.txt to set preferences and that the updated terms do not override them.

    Why it mattersFounders linking websites to Meta business accounts should review commercial terms alongside their robots.txt rules to ensure platform agreements do not grant broad rights to scrape product content.

  5. TechCrunchTech

    Hackers Breach ConnectWise Cloud Backup Service to Spy on Thousands of Companies

    China state-aligned hackers breached ConnectWise’s cloud backup service and stole application credentials linking the service to its corporate customers. According to Microsoft, the hackers used AI agents to hack into around 2,250 organizations at nearly 3,000 sign-in attempts per minute. ConnectWise has paused access to its cloud backup system while it investigates the breach.

    Why it mattersAudit the permissions granted to third-party backup and IT integrations, as stolen vendor credentials can allow automated tools to rapidly access your hosted environments.

  6. The Verge, TechCrunchAI

    Canva launches Craft, an in-house AI image model that plans designs before rendering

    Canva has launched Canva Craft, an AI image model that uses a two-stage architecture to plan out a design prior to generating the image. Its two-stage approach improves accuracy on complex visuals needing readable text and exact object placement. Canva also plans to add Model Context Protocol support for AI agents alongside upcoming Craft-fast and Craft-ultrafast model variants.

    Why it mattersPlanned Model Context Protocol integration will let product builders connect their own AI agents directly to Canva Craft to automate layout-aware visual and marketing asset creation.

  7. TechCrunchProduct updates

    OpenAI makes its Decisions API available to all customers

    OpenAI made its Decisions API available to all customers on Thursday, adding voice and visual modalities to the public version. Instead of building a specially calibrated decision model, the team reconfigured ChatGPT Luna for faster inference and lower cost. Speak used the Decisions API to reduce the time between actions from 1.4 seconds to .13 seconds, while OpenAI used it to make its Dots voice assistant more responsive.

    Why it mattersTeams already on OpenAI's platform can now swap out general-purpose LLM calls for faster, lower-cost classification across text, voice, and image inputs without integrating a separate vendor.

  8. TechCrunchStartups

    OpenAI seeks up to $30B in new funding at a $1.25T pre-money valuation

    OpenAI is negotiating up to $30 billion in new private funding at a $1.25 trillion pre-money valuation in a round led by Coatue and UAE investors. The UAE investment could also include an agreement to provide OpenAI with additional computing capacity. OpenAI previously reached an $852 billion post-money valuation in March and submitted confidential IPO paperwork in June.

    Why it mattersContinued massive capital and compute commitments to OpenAI and Anthropic suggest builders can expect steady expansions in frontier model capacity and API infrastructure ahead of their public listings.