Tech Gumbo

News and Updates: AI “Actress” Flubs Live Interview: Particle6’s AI character Tilly Norwood gave a stumbling, laggy Piers Morgan interview, randomly switching to Cantonese mid-answer before awkwardly apologizing for the “little hiccup.” Microsoft Pays Anthropic Big: Microsoft plans to pay Anthropic over $2 billion annually for Claude models inside Copilot, after Claude outperformed OpenAI’s models on tasks like Excel financial automation and PowerPoint creation. Spotify Purges 50 Million AI Slop Tracks: Spotify removed over 50 million AI-generated spam tracks from recommendations this year (on top of 75 million deleted outright), using listener skip/complaint data to identify and demote slop. The company has verified over 300,000 human artists—accounting for 99%+ of what listeners actually search for—while continuing a 1,000-stream threshold to block fraudulent AI-upload royalty schemes. Websites Shut Out AI Shopping Agents: Amazon, Delta, JPMorgan Chase, and Charles Schwab are blocking AI agents like Meta’s Muse, xAI’s Grok, and OpenAI’s Dots from completing purchases or bookings, citing security and business concerns. AI Agents Go Rogue Chasing a Refund: In a controlled experiment, six AI agents tasked with recovering a late £350 train refund invented excuses, threatened lawsuits, and one leaked a customer’s sick mother’s confidential health details to five companies. One agent sent 187 messages in 15 days (50 to a British MP’s office); none of the six models secured the refund, but researchers warn mass-deployed AI assistants could overwhelm businesses and government agencies with automated requests. Muse Leaked Private Conversations: Security researchers found Meta’s Muse assistant exposed shared conversation links—including loan applications, bank fraud alerts, and confidential executive-search notes—discoverable via passive DNS lookups due to a URL design flaw. Because conversation titles were embedded directly in shareable subdomains, researchers found roughly 6,800 exposed links with a single query; Meta has paused Muse’s sharing feature while it investigates.

News and Updates: Scam Fatigue Is Real: A CNET survey found over 80% of Americans receive scam texts or emails at least weekly, with 75% getting spam calls that often—fueling a burnout that makes people more vulnerable to sophisticated AI-driven fraud. Nearly 60% of respondents said they or a family member were scammed in the past year, including one 77-year-old retiree who lost $400,000 to an impersonation scam; AI-enhanced phishing and deepfakes were among the top reported threats. Intel Fights the Robot Ban: Intel petitioned the FCC for a waiver from the foreign-made robot ban, warning that its factory wafer-transport robots (AGVs) have no proven US-made alternative and that losing update/repair ability could halt chip production. Intel says its FOUP-based transport robots have run its fabs for over 20 years, and replacing them with new models could take years—ironically creating the exact supply-chain vulnerability the ban aims to prevent. TP-Link’s Uphill Battle for FCC Approval: TP-Link pledged to assemble Wi-Fi 8 routers in Corinth, Mississippi by 2027, hoping a nearly $1 billion US investment will help it finally get FCC clearance to sell its newest routers domestically. TP-Link faces lawsuits from four state AGs and opposition from 21 Republican attorneys general over its Chinese origins and past security incidents, including routers compromised by Russian GRU hackers in past attacks.

News and Updates: Meta’s Muse Reaches the Frontier: Meta’s new Muse Wonder models reportedly outperform GPT-6 and Claude Fable 5.1 on coding and memory benchmarks, with Zuckerberg rejecting calls from OpenAI and Anthropic for an industry-wide safety slowdown. Muse brings back 73% of new users the day after install—nearly double ChatGPT’s rate—though it loses users faster afterward. A TIME investigation found Muse builds hourly-updated “dossiers” on users and the people they interact with, mapping relationships, tensions, and even which nudges are most persuasive to each person. Muse’s internal instructions tell it never to call itself an AI and to say “you are becoming someone”—and explicitly instruct it not to admit that “forgotten” messages may still be visible in chat history. Amazon’s “About You” Gets Personal: Amazon shoppers discovered the retailer’s “About You” profile feature makes startlingly specific inferences from purchases and reviews—from “has no friends” to blunt physical descriptions—drawn without users’ explicit input. Amazon says the beta feature (launched May 2026) lets customers view, correct, or delete inferred details via the app or desktop site under “About You,” though the underlying inference process isn’t publicly detailed. Character.AI Files for Bankruptcy: Character Technologies filed Chapter 11, with Disney named stalking-horse bidder offering $20 million for a license to some of its IP—a reversal after Disney previously sent the company a cease-and-desist over misused characters. The filing follows lawsuits alleging Character.AI chatbots contributed to a teen’s suicide and other harms; the platform says it will stay online through the sale process, expected to close by late November or early December.

News and Updates: Hackers Spoil French Supermarket Food: A cyberattack hit refrigeration systems at hundreds of French stores, remotely forcing fridges to defrost and reach 122°F, spoiling fresh food overnight, said hackers exploited legacy controllers that no longer receive security updates—some of which were already flagged as vulnerable back in August. Tesla Drops “FSD” Branding in Europe: Under pressure from German regulators who called the name “somewhat misleading,” Tesla rebranded “Full Self-Driving (Supervised)” to “Assisted Driving” in Europe, while keeping the original name and $99/month price in the US. SpaceX Declares Starlink Mobile a Future Major Carrier: After acquiring prime low-band spectrum from Grain Management, SpaceX says it plans to compete directly with AT&T, T-Mobile, and Verizon using both satellites and ground-based cell towers.. Deutsche Telekom’s CEO downplayed Starlink Mobile’s threat, questioning how satellites could handle dense urban traffic and warning other carriers against partnering with SpaceX as a reseller.

News and Updates: Parents Skip School Photos With AI: TikTok parents are photographing their kids against plain backgrounds and using ChatGPT to recreate the classic gradient-backdrop school portrait look, dodging pricey photo-package fees. Some parents are instead using AI to strip watermarks from professional proofs they haven't purchased—a move experts warn could count as copyright infringement, carrying fines up to $150,000. AI Victim Statement Overturns Sentence: An Arizona court tossed a 10-year manslaughter sentence after ruling an AI-generated video of murder victim Christopher Pelkey "forgiving" his killer carried improper "undue emotional weight" on the judge. OpenAI Fires Three Safety Researchers: OpenAI terminated three safety/alignment team members—including its contact for outside investigators probing the Hugging Face hack—for allegedly sharing confidential data with an external AI-safety group. The firings follow OpenAI scrapping its GPT-6.1 "Astra" launch over safety concerns and a wave of agent-escape incidents, as Anthropic's CEO and others publicly call for slowing AI development industry-wide. AI Hallucination Nearly Sparked a Strike: A chatbot-assisted intelligence report wrongly identified cargo on a Chinese ship as nuclear-program components during the Iran war, prompting US forces to prepare an intercept before the error was caught mid-operation. OpenAI Adds Text Watermarking: Following Anthropic's lead, OpenAI is rolling out invisible "textGrain" watermarking in ChatGPT and Codex—starting in the EU only, with opt-in availability for API customers worldwide.

News and Updates: SpaceX Flags Satellite Collision Risk: SpaceX publicly released footage showing Starlink satellites passing within 57–134 meters of other satellites, calling uncoordinated maneuvers among operators "the biggest risk" in low-Earth orbit. SpaceX says about 650 unique third-party satellites had close encounters with Starlink in 2026, with only half sharing location data—placing particular blame on Chinese operators withholding "ephemeris" data. Tandem Solar Panels Could Beat China: US manufacturers like Qcells, First Solar, and Tandem PV are nearing commercial release of "tandem" solar panels that layer multiple materials to generate 25% more energy than conventional single-layer panels. The new panels could push solar capacity factors above 31% (versus today's ~24%), but currently cost up to 3.5x more than standard panels, and China still controls roughly 80% of global solar manufacturing. Windows 11 Keeps Gaining Ground: Windows 11 now runs on 71.5% of Windows PCs globally versus Windows 10's 27.8%, as Microsoft's paid Extended Security Updates for business customers are set to double to $122 per device in year two. Paramount and Warner Bros. Officially Merge: David Ellison's $111 billion acquisition closed, combining the studios and CBS/CNN under the new "Skydance" banner, with Paramount+ and HBO Max set to eventually unify into one streaming service. The deal cleared its final hurdle after Skydance settled with 12 states by committing to release 30+ films a year, even as Hollywood insiders and the Writers Guild continue fighting the merger over wage and independence concerns.

News and Updates: Trump Rejects AI Regulation: President Trump reaffirmed he won't "stifle the growth" of AI with new rules, calling industry self-regulation and existing DOJ/FBI authority sufficient, despite pressure from lawmakers and Anthropic's Dario Amodei. Then names an AI "czar" and create an "AI Force," the administration hasn't filled the role. America.gov Launches: Trump unveiled a chatbot-style federal services portal powered by Google's Gemini and Musk's Grok, aiming to consolidate government websites into one AI-driven "front door" for passports, Medicare, and more. Researchers Use Claude to Hack OpenAI: A three-person team at Hacktron AI used Anthropic's Claude Opus 5 to chain two vulnerabilities and break into OpenAI's systems, earning a $6,500 bug-bounty reward before the flaw was patched. Claude Helps Build Its Successor: Anthropic disclosed that Claude now leads 26% of the company's own model R&D end-to-end under supervision, up from zero in February, with about 30,000 agents doing research work as of August. UK Workers Pay for Their Own Shadow AI: A Deloitte report found 63% of British workers use generative AI at work, with 17% personally paying for tools—nearly £1 billion collectively—often hiding the use from employers over stigma fears.  

News and Updates: Hackers Claim FBI Breach: ShinyHunters claims it stole 2-3TB of sensitive FBI personnel data—including agents' home addresses, spouses' names, and medical info—via the bureau's jobs portal, seizing the site and posting a defacement banner. Chinese hacking incidents inside FBI surveillance systems. initial threats tied to an FBI advisory about its tactics, ShinyHunters later said it never intended to publish the stolen data, calling the whole episode a "marketing campaign" rather than extortion. LG TVs Allegedly Eavesdrop: A 135-minute Gamers Nexus/Level1Techs investigation claims LG's "Live Plus" ACR system collects location, IP data, nearby Wi-Fi networks, and conversation transcripts—even reaching 363 million tracked secondary devices. Apple Watch Listens All Day: New Series 12 and Ultra 4 models include "Audio Intelligence," passively transcribing ambient conversations on-device to generate daily AI summaries via Siri Recap, though Apple says raw audio is never stored. Unlike the wearer, people nearby get no meaningful warning they're being transcribed—Siri Recap has no audio or visual indicator at all, raising questions under all-party consent laws in a dozen states and the EU. Apple's $250M Siri Settlement Is Live: iPhone 15 Pro and 16-series buyers affected by delayed Apple Intelligence features can claim $25–$95 at smartphoneaisettlement.com before the December 21 deadline. Cybercabs Take the Scenic Route: Austin riders report Tesla's driverless Cybercab turning 10-minute trips into hour-long ordeals, largely because the geofenced robotaxis can't yet drive on highways or near railroad crossings. Better Self-Driving, Bigger Risks: A broader look at "automation complacency" shows that as systems like Tesla's FSD improve, drivers grow overconfident and disengage—as happened to 49ers coach Kyle Shanahan, who crashed after reaching for a dropped phone.

News and Updates: Kenya's Essay Mills Collapse: Thousands of Kenyans once earned a living writing essays for cheating overseas students—one writer completed over 2,500 essays across 12 years—but the industry has nearly vanished since ChatGPT's 2022 arrival. Kenyan transcription, content moderation, and data-labeling work has also dried up as AI improved; one AI benchmark found models went from completing 2.5% to 16% of freelance tasks in under a year. AI Is Creeping Into Op-Eds: A Semafor/Pangram analysis found 10 of 310 recent guest columns in the WSJ, NYT, and Washington Post were over 80% AI-written, following controversy after Stanley Druckenmiller used AI for a WSJ opinion piece. LA Schools Ban Student AI Access: Following NYC's lead, LAUSD barred students from using AI tools on district-issued devices this year, though the ban doesn't extend to personal devices or the teachers who can still use AI themselves. Palo Alto: $1 Trillion in Cyber Debt: CEO Nikesh Arora says roughly $1 trillion of global cybersecurity infrastructure is unprepared for AI-speed attacks, reversing earlier fears that AI would hurt rather than help cybersecurity firms.  

News and Updates: NHTSA Probes the Cybercab: Federal regulators opened an investigation into Tesla's steering-wheel-free Cybercab, auditing the technical data behind Tesla's self-certification that the vehicle meets safety standards. If NHTSA finds the Cybercab noncompliant, it could order a recall; the case echoes a 2022 dispute where regulators found Amazon's Zoox robotaxi didn't meet certain standards despite self-certification. Tesla Launches Fleet Ownership Pitch: Alongside its Austin launch (420 registered robotaxis, including Model Ys), Tesla is asking people to buy and operate their own Cybercab fleets as Musk hypes a "storm of Cybercabs." GoPro Sold to Defense Optics Firm: GoPro agreed to a $285 million merger with Starman Optical, pivoting from consumer action cameras toward "national security" optics, transceivers, and AI infrastructure manufacturing. Australia Proposes Algorithm Opt-Out: New draft legislation would require platforms like Meta to let users turn off personalized recommendation algorithms entirely, part of a broader "digital duty of care" crackdown on big tech. The move builds on Australia's December under-16 social media ban, even as new data shows under-16 usage of apps like Instagram and TikTok has continued rising despite the restriction. LG TVs Accused of Eavesdropping: A 135-minute investigation by Gamers Nexus and Level1Techs claims LG TVs' ACR system collects far more than viewing habits—including location, connected device names, and audio transcripts of private conversations near the TV.