AI wanted your workplace archives, your bank transactions, and a spot on your face this week. Getting consumers to let it book a ride proved harder. Meanwhile, Nvidia faced a leadership overhaul, Amazon gave surveillance cameras mobility, and security teams got another reminder that the future still ships with default passwords.
Here’s everything you missed this week in the Daily Tech Insider newsletter.
Top news
Nvidia faces a historic leadership transition
Jensen Huang has left Nvidia after 33 years, stepping down from his roles as CEO, president, and director. Chief Financial Officer Colette Kress is serving as interim CEO while the board evaluates both internal and external candidates for a permanent successor.
AI agents still struggle to win consumers
Despite intense investment in agentic commerce, third-party AI shopping agents account for only a tiny share of Uber and DoorDash transactions. Travel platforms are reporting similarly limited adoption, and some companies are restricting outside agents rather than opening their services to automated intermediaries. These figures exclude in-house AI tools, and DoorDash’s developer connector remains in restricted beta.
Workplace archives become AI training material
AI developers are simultaneously looking for more specialized training data. Data brokers are offering companies as much as $2 million for workplace archives, including emails, Slack messages, and other business files that could be used to train models. The practice raises unresolved questions about employee consent, re-identification, confidentiality, and ownership of workplace communications.
Robots return to homes and businesses
Ring unveiled two mobile security cameras: the $249.99 indoor Drone Cam for homes and the $999.99 Rover Cam for businesses. Both rely on Ring’s cloud-assisted AI Pilot and internet access; scheduled patrols and saved video require compatible subscriptions. Rover is due Dec. 16, while Drone Cam targets Jan. 27, 2027, pending FCC authorization.
Amazon is making a separate attempt to revive its home-robot ambitions. The company is assembling a team to develop a new robot capable of navigating homes, including members of the original Astro group. Its design, capabilities, price, and release plans remain undisclosed.
Insider intel
AI moves onto local devices and into everyday services
Apple subsidiary Claris is privately testing Harbor, an application for running open-weight AI models on Apple silicon. The software is also designed to pool memory across compatible Macs, potentially making larger local models more practical. Apple has not disclosed pricing, performance, compatibility requirements, or a public launch date.
Meta launched Muse Ember, a compact model intended to reduce the cost of repetitive agent tasks such as document extraction, support-ticket routing, and coding subtasks. It supports a one-million-token context window, with prices beginning at $0.10 per million input tokens and $0.50 per million output tokens. Those rates triple for inputs exceeding 200,000 tokens.
Samsung may be preparing to turn some mobile AI features into a subscription business. Strings found in an application indicate that Galaxy AI Premium could cost $4.99 per month or $47.99 annually, offering additional AI tools and higher usage allowances. Samsung has not formally announced the service or provided a launch date.
Anthropic is also expanding AI into personal finance. A US beta gives Claude Pro, Max, and Ultra subscribers read-only financial features through Plaid. Users can connect bank and credit card accounts so Claude can analyze balances, transactions, duplicate charges, and spending patterns, but the assistant cannot move money.
Apple retires a long-running support resource
Apple is closing its Support Community, ending a user-to-user help forum and taking its archives offline. The company stopped accepting new questions and replies on Nov. 10 and plans to remove the community completely on Nov. 20.
Security alerts 1
Actively exploited flaws demand immediate updates
Google patched CVE-2026-125350, a high-severity XML memory flaw in Chrome. It is the eighth Chrome zero-day known to have been actively exploited in 2026. The desktop update addresses nine vulnerabilities in total, and users must install it and relaunch Chrome to complete the process.
Application and remote-management flaws expose organizations
ScreenConnect attacks installed rogue remote clients and SCKit malware on at least 288 endpoints across 48 organizations. Researchers connected the activity to a flaw in RMM+ versions 1.0.31 and 1.0.32 that accepted a blank access key when an administrator had never configured one.
An iOS sandbox-bypass vulnerability known as worse_query could allow a malicious installed application to access other apps’ files. The exposed information could include wallet data and login details, although access does not automatically defeat iOS Keychain encryption. No attacks or thefts exploiting the flaw have been confirmed, and Apple has not explicitly confirmed that iOS 27.2.1 or 26.7.3 fixes it.
Researchers also demonstrated how weak configuration practices can create physical risks. By combining a firewall bypass with default credentials, they compromised ForgeRock identity systems at two unnamed automakers. The access enabled account takeovers and the remote unlocking, starting, and tracking of connected vehicles. Both automakers patched the problem following private disclosure.
Event lures target work accounts
Another campaign is exploiting interest in high-profile technology companies. Attackers are impersonating OpenAI, SpaceX, and Anthropic events to steal Microsoft and Google work accounts. Targets are persuaded to approve Microsoft device codes or hand over Google passwords and session cookies. Users should independently verify invitations, while administrators should restrict device-code authentication and deploy phishing-resistant multifactor authentication.
Security alerts 2
Device and application updates address security flaws
Microsoft addressed 1,235 vulnerabilities in its November security updates. That tally includes Chromium-based Edge flaws. The fixes include CVE-2026-124194, an actively exploited Outlook vulnerability that can expose NTLM hashes when a user opens a malicious message file. Apply the applicable Office updates separately; installing Windows updates alone does not necessarily fix Outlook.
Apple released security updates across its device portfolio, covering iPhones, iPads, Macs, Apple Watches, and Vision Pro hardware. The flagship iPhone update fixes nine vulnerabilities, including flaws affecting iMessage and Bluetooth. Older iPhone XS, XS Max, and XR models are receiving iOS 18.7.12.
AI applications and public sharing create data risks
Security company Claroc found more than 35,000 Grok-built applications publicly accessible. Most were harmless calculators or games, but some exposed patient records, plaintext passwords, and proprietary business information. Randomized links did not provide meaningful protection, and even a Google sign-in prompt placed on a test application failed to stop unauthorized access.
Apple is introducing a new policy for publicly shared cloud content. Beginning with iOS 27.3 and related updates, Apple plans to scan iCloud files shared through the “Anyone with the Link” setting for harmful material. Private, unshared files will be excluded, but public links do not use end-to-end encryption. Flagged content could be deleted and may lead to account penalties.
Deepfake voices target payment approvals
Prosecutors allege that criminals used deepfake voices impersonating CEOs and lawyers to steal more than $129 million. The cloned voices were allegedly used to request secret wire transfers. Organizations should verify urgent payment instructions through trusted, independent channels rather than relying on the incoming call or message.
Industry shakeups
Anthropic prepares for a potentially enormous public debut
Anthropic is targeting a fully diluted valuation of $1.4 trillion in its planned initial public offering. The company set a price range of $70 to $78 per share ahead of an expected Nov. 19 Nasdaq debut. Its public offering and accompanying private placement could raise tens of billions of dollars.
Nvidia investigates espionage claims amid leadership turmoil
Nvidia’s board launched an independent inquiry into reports that Jensen Huang was targeted by a foreign intelligence operation. Three House committee chairs requested a closed-door briefing on the matter. Nvidia said it had found no indication that company secrets or national security had been compromised.
SpaceX and Qualcomm plan a nationwide 5G buildout
SpaceX selected Qualcomm to support Starlink Mobile’s planned US 5G network. Qualcomm will provide Dragonwing processors and baseband software for the project. Construction is expected to begin in 2027, with the network ultimately intended to span tens of thousands of cell sites.
New AI hardware targets data centers and wearables
Google announced the TPU 8i Max and TPU 8tx as additions to its eighth-generation accelerator portfolio. The TPU 8i Max emphasizes memory-intensive workloads, while the TPU 8tx focuses on security. The systems are expected in 2027, potentially giving enterprise AI buyers more specialized alternatives to Nvidia hardware.
At the consumer end of the market, Google and Samsung opened preorders for Android XR smart glasses ahead of a Dec. 10 launch. Warby Parker models start at $395, while Gentle Monster frames begin at $449. The camera-and-audio glasses use Gemini to provide translation, directions, and other forms of assistance.
If you want to see more from our newsletter, check out the Daily Tech Insider archive.
Discover more from PressNewsAgency
Subscribe to get the latest posts sent to your email.