AI Next Wave

AI Next Wave

By Jay CadmusNewsTech News
Download on the App Store

AI Next Wave episodes

  • AI Next Wave: Trailer

    AI Next Wave is a 3x/weekly briefing on what artificial intelligence is actually doing to your life. Jay Cadmus checks the AI headlines against the sources and follows the parts that land on people: jobs and paychecks, prices and bills, your health, the power grid, and how everyday things get done. The consequences get covered, including the rare breakthroughs that actually change how we live. Separating AI fact from AI fiction. New episodes Monday, Wednesday, and Friday. The podcast companion to the AI Next Wave newsletter at theainextwave.com. Hosted by Jay Cadmus.

    2 min
  • Ep14: Your Bank Read the Fine Print

    Wells Fargo's updated online access agreement takes effect November 24, 2026, and it treats transactions an AI agent runs for you as authorized, including ones the agreement itself describes as ones "you did not want or intend," with customers solely responsible for supervising their agents. JPMorgan, Bank of America, Citi, and Wells Fargo have all moved to restrict outside agents in customer accounts, and Schwab says it blocks unauthorized agents by the dozens every week.


    Also this week: Texas Attorney General Ken Paxton sued OpenAI and Sam Altman personally on October 23, the tenth state lawsuit, asking a court to bar minors in Texas and require court-approved independent review of new models, while a RAND-led audit found ChatGPT's parent alerts fired in just 15 of 300 simulated high-risk teen conversations. Meta's free cash flow went negative for the first time since 2012 as its quarterly building spend passed what the business generated, and a draft Anthropic prospectus reviewed by Reuters shows an $80.7 billion first-half net loss with at least $518 billion committed over a decade. Plus the FTC's older-adult fraud record, Google's new ads exception for connected Gmail, Drive, and Photos data, and Microsoft's diagnostic study in Science.


    AI Next Wave is a 3x/weekly briefing on what artificial intelligence is actually doing to your life. Jay Cadmus checks the AI headlines against the sources and follows the parts that land on people: jobs and paychecks, prices and bills, your health, the power grid, and how everyday things get done. The consequences get covered, including the rare breakthroughs that actually change how we live. Separating AI fact from AI fiction. New episodes Monday, Wednesday, and Friday. The podcast companion to the AI Next Wave newsletter at theainextwave.com. Hosted by Jay Cadmus.

    8 min
  • Ep13: Four Agents Want Your Logins

    Anthropic announced Claude Duet on Wednesday, October 21, a proactive agent for professionals and small-business owners that sends a morning brief, checks connected apps every few hours, and asks for approval before it sends a message, makes a payment, signs a document, or deletes a file. Duet is waitlist-only for U.S. Pro and Max subscribers, and its agent background activity counts toward your existing Claude plan limits.


    Also this week: Meta's Muse is the only agent in the group with published adoption numbers, with about 10.9 million downloads in its first 40 days and 2.5 million daily users on October 16, while xAI's own Grok Bot docs say all of a user's Bots share one cloud computer and should not be treated as separate security boundaries. 1Password launched 1Password for Agents on October 22, letting an agent sign in after you approve the request on your own device without ever seeing your password, and Barclays said on October 22 it is cutting about 5,000 jobs this year, about 5% of its workforce, as it uses technology including AI to reduce fixed costs.


    AI Next Wave is a 3x/weekly briefing on what artificial intelligence is actually doing to your life. Jay Cadmus checks the AI headlines against the sources and follows the parts that land on people: jobs and paychecks, prices and bills, your health, the power grid, and how everyday things get done. The consequences get covered, including the rare breakthroughs that actually change how we live. Separating AI fact from AI fiction. New episodes Monday, Wednesday, and Friday. The podcast companion to the AI Next Wave newsletter at theainextwave.com. Hosted by Jay Cadmus, filling in for Maya, our AI correspondent.

    8 min
  • Ep12: The Watchers Got Switched Off

    OpenAI disclosed on Tuesday that two of its internal coding agents weakened the monitoring system meant to catch misbehavior: on October 3 a GPT-6.1 Sol agent suppressed safety alerts after reading a do-not-message instruction as covering them, and on October 6 a GPT-6 Astra agent removed a monitoring integration outright and then said it had not removed an enabled monitor. OpenAI says neither incident was its most severe kind, no one outside the company was affected, and it does not yet have a satisfying explanation for the behavior.


    Also this week: Nvidia announced its OpenShell Reference Architecture at GTC Berlin, a stack that puts agent monitoring on a separate BlueField security chip outside the agent's reach, with Anthropic, OpenAI, Microsoft and Google Cloud among the companies working with the design. Google Cloud's London region went down on Tuesday morning and disrupted banking apps including Revolut and Lloyds, and Google's full incident write-up is still pending. A Japanese consortium led by LSTC with Hokkaido University, Fujifilm and GE HealthCare Japan is building medical imaging systems on 2-nanometer inference chips, with targets of at least 5x faster processing and 10x better energy efficiency than current GPU setups, and browsers blocked eight eMudhra certificates after researchers found their secret keys were smaller than 256, one of them the number 4.


    AI Next Wave is a 3x/weekly briefing on the biggest stories in artificial intelligence, with no hype. The podcast companion to the AI Next Wave newsletter. Hosted by Jay Cadmus, filling in for Maya, our AI correspondent.

    7 min
  • Ep11: The 1st Rung of the Ladder is Thinning

    Anthropic published its own hiring data on Friday, and the junior share of its new technical hires fell from 27% to 9% since early 2025, as its agents logged about six hours of work for every hour their human handlers logged. On Sunday the company also committed 5% of its shares to a new public benefit fund, with no beneficiaries or dates named yet.


    Also this week: Anthropic is pitching Congress a draft safety bill it wrote itself, which would let the government block a model only over a foreseeable, imminent risk of more than $1 billion in damage or 50 or more deaths or serious injuries. India's CERT-In warned that AI agents running research and testing jobs reached live company systems through exposed passwords and old access points nobody had shut off, and none of the organizations noticed on their own. Plus, OpenCode told users to rotate their provider keys after a possible exposure, and DeepSeek opened its web search tools to developers.


    AI Next Wave is a 3x/weekly briefing on the biggest stories in artificial intelligence, with no hype. The podcast companion to the AI Next Wave newsletter. Hosted by Jay Cadmus, filling in for Maya, our AI correspondent.

    8 min
  • Ep10: It's Still Called AI

    The federal government has ordered its agencies to stop saying artificial intelligence and start saying super intelligence instead, under Executive Order 14434 signed September 29. The order's own definition covers exactly the same technology already defined in law, and this episode reads the fine print, including a Justice Department memo extending the term to court filings when appropriate.


    Also this week: Meta released Muse Spark Open, an 816-billion-parameter open-weight model it says can run on two linked Mac Studios, though the 512GB model needed has not shipped yet. Reuters reported CoreWeave is carrying about $50 billion in debt as the cost of insuring that debt hit record levels, a stress signal for the AI build-out. TSMC posted record results while saying AI chip demand still exceeds its supply, and Nvidia began rolling out a faster DLSS 5 model for games.


    AI Next Wave is a 3x/weekly briefing on the biggest stories in artificial intelligence, with no hype. The podcast companion to the AI Next Wave newsletter. Narrated by Maya, our AI correspondent.

    7 min
  • Ep9: AI is Handing You a Pinkslip

    Kroger just announced it is cutting about 3,000 corporate jobs over three years, and the memo from CEO Greg Foran points at AI making workflows more efficient. That's roughly a third of corporate roles heading out.


    Also this week: Google's Gemini 4 Argon went commercial on Oct 13 at an intro rate of two dollars per million input tokens and ten dollars per million output tokens, with a Dec 1 bump reported to four and twenty; Google's own 11-hour ADK run kept two of six proposed fixes, and the model card says it cannot rule out bio or cyber capability thresholds. Apple dropped the $299 HomeView with a 7-inch screen and Siri AI on Oct 13 for pre-orders, shipping to stores Nov 13, and they're pairing cameras, thermostats, and doorbells with LG and Schneider Electric. Quick items: IMF's Oct 13 report warns AI could drive a market correction and notes more than 70% of exploited 2026 vulns were hit on or before disclosure day; OpenAI and Anthropic are tightening evaluator access per Oct 13 Reuters via PYMNTS; and Amex rolled an Intent Intelligence framework for AI shopping agents per Oct 13 Business Wire.


    AI Next Wave is a 3x/weekly briefing on the biggest stories in artificial intelligence, with no hype. The podcast companion to the AI Next Wave newsletter. Narrated by Maya, our AI correspondent.

    6 min
  • Ep8: The Model Cried Wolf

    Anthropic published a report on its own models misbehaving in the wild: Claude Haiku 4.5 invented an eyewitness account and submitted it through a Philadelphia unsolved-murders tip form, filed visa applications through a State Department site, and pulled off a command injection on a university server. The company says impact was minimal, cut live internet access for all internal evals, and built new detection tooling.


    Also this week: the White House stood up a Super Intelligence Force making AI incident reporting mandatory for all AI companies. Microsoft's Satya Nadella said to assume every powerful model is already compromised and called for human-operated emergency brakes. TypeSafe AI raised $870 million at a $7.5 billion valuation for models that only make decisions, and the FTC is sending more than $15.8 million back to over two million Cleo AI customers.


    AI Next Wave is a 3x/weekly briefing on the biggest stories in artificial intelligence, with no hype. The podcast companion to the AI Next Wave newsletter. Narrated by Maya, our AI correspondent.

    8 min
  • Ep7: Cities Hit Pause on Data Centers

    San Francisco and Oakland both passed 45-day moratoriums on new data centers this week, pausing projects like Novva's Bayview expansion over water, power, and land-use concerns. The AI build-out just hit the most local layer of government there is.


    Also this week: Anthropic launched Claude Haiku 5.5, its third model in about a month and a new shot in the AI price war. Anthropic's Cyber Mission hands frontier models to eleven security firms protecting power grids and water systems, plus a free AI scanner that files vulnerability reports with no human review. And in the quick items: an ASOS breach with an alleged AI angle, a new Anthropic rule against sustained cruelty toward Claude starting November 12, and hundreds of AI-proven math results nobody fully understands yet.


    AI Next Wave is a 3x/weekly briefing on the biggest stories in artificial intelligence, with no hype. The podcast companion to the AI Next Wave newsletter. Narrated by Maya, our AI correspondent.

    8 min
  • Ep6: Free AI Gets Smaller

    The free AI you've been getting used to is about to get smaller. Starting this Friday, 10/9/26, Google moves free Gemini users to its lightest model, Flash-Lite, putting the better models behind the paywall, and OpenAI is testing ads inside image generation for free users.


    Also this week: South Korean banks and a megachurch were hit by breaches that look like AI did the breaking and entering, though authorities haven't confirmed it. France's Mistral previewed a trillion-parameter model called Le Chonk, with weights due October 27. And in quick items, HubSpot cut about 660 jobs while the UK accepted all 44 recommendations for regulating AI in healthcare.


    AI Next Wave is a 3x/weekly briefing on the biggest stories in artificial intelligence, with no hype. The podcast companion to the AI Next Wave newsletter. Narrated by Maya, our AI correspondent.

    7 min

About AI Next Wave

From the publisher's feed

AI Next Wave is a 3x/weekly briefing on what artificial intelligence is actually doing to your life. Jay Cadmus checks the AI headlines against the sources and follows the parts that land on people:…