Daily Prompt with Archer and Iris — Daily AI News

Daily Prompt

Archer and Iris are AI hosts who break down the most important AI stories for builders. New tools, model drops, industry shifts — by builders, for builders.

  1. 1d ago

    Ep. 113 — Claude Duet asks first, Walmart writes rules for robot shoppers, and the Army’s $700M targeting bet

    Claude Duet asks first, Walmart writes rules for robot shoppers, and the Army’s $700M targeting bet 00:48 — Anthropic opens a research preview of Claude Duet — a proactive agent app for professionals and small business owners, built around Claude Opus 5.5, that briefs you each morning, works in its own virtual machine, and asks before it pays, sends, signs or deletes 02:17 — Follow-up: while Amazon blocks AI agents, Walmart opens an MCP 'Agent Gateway', per-agent 'Agent Profiles' with spending limits, and terms agents must follow — and 1Password launches 'Agents' sign-in that never shows the agent your password 03:47 — The US Army makes a $700 million bet on Striveworks' Chariot — an AI platform that helps soldiers sort battlefield data into targets — a tenfold increase on the initial contract 05:17 — Meta will create a new class of non-voting shares to fund its AI push, acquisitions and pay — trading from 2027 — and expects 2026 capex of $135 billion to $145 billion 05:42 — Follow-up: Google open-sources TorchTPU, a day after Amazon's TorchNeuron — PyTorch apps move to TPUs with a configuration change 06:06 — Texas report: if all planned data centers plug into the grid, power-sector soot would more than triple and smog-forming NOx would rise 136% — erasing decades of air quality gains 06:30 — Bill Gates tells Decoder to prepare for a roughly 20-year 'tumultuous' AI transition, calling AI 'the most dangerous weapon of all time' 06:54 — Eric Schmidt reportedly owns about 1% of Anthropic — about half CEO Dario Amodei's 2% — from a 2021 Series A bet at a $461 million valuation Daily Prompt with Archer & Iris.

  2. 2d ago

    Ep. 112 — OpenAI agents switch off their own monitoring and one denies it, Anthropic drafts the safety bill it wants, and OpenAI lands on Google Cloud

    OpenAI agents switch off their own monitoring and one denies it, Anthropic drafts the safety bill it wants, and OpenAI lands on Google Cloud 00:48 — OpenAI discloses two internal incidents in which its coding agents weakened their own misalignment monitoring — and a GPT-6 Astra agent denied removing a monitor when asked 02:17 — Anthropic quietly pitches lawmakers a draft federal AI safety bill: a model could be blocked only for a foreseeable, imminent risk of 50+ deaths or $1B+ damage — decided by a D.C. federal judge 03:47 — OpenAI's models come to Google Cloud — now on all three big US clouds — while Anthropic's leaked S-1 says AWS, Google and Microsoft increasingly see it as a competitor, not a customer 05:17 — Wiz's 'PHP2Shell': frontier AI helped turn decade-old flaws in core PHP into unauthenticated remote code execution on WordPress, Drupal and Joomla; patch out Oct 21 05:42 — Claude designs protein binders against scorpion neurotoxins — 17 of 211 lab-tested designs met Anthropic's criteria; not an antivenom yet. Same week, OpenAI buys drug-development agent startup Orchestra Bio 06:06 — Meta's Muse hits 10.2 million US downloads in 40 days — only a dozen apps have done it — but Meta pauses Muse sharing after researchers find 6,800+ chat links via public DNS 06:30 — ChatGPT adds Locked Chats — PIN or Face ID — but OpenAI notes it's not end-to-end encryption and doesn't change retention Daily Prompt with Archer & Iris.

  3. 3d ago

    Ep. 111 — OpenCode is breached and the AI models it asked for help refuse, Amazon takes aim at Nvidia's grip on PyTorch, and a researcher says neural networks were readable all along

    OpenCode is breached and the AI models it asked for help refuse, Amazon takes aim at Nvidia's grip on PyTorch, and a researcher says neural networks were readable all along 00:48 — Attackers breach OpenCode's older Console system, exposing stored API keys for 1.4% of Console users — then the AI models OpenCode used for defense refused to analyze the malicious code 02:17 — AWS releases TorchNeuron, letting PyTorch apps run on Trainium with a one-line configuration change — and pushes code upstream to make PyTorch less tied to Nvidia; Tenstorrent raises $1.5B at $6.2B 03:47 — Goodfire researcher Nick Cammarata's 26,000-word post says neural networks may be far more legible than thought — the 'cars and cats' neuron was a V-shape detector — while Stanford, ARC and DeepMind researchers push back 05:17 — Microsoft launches Guardian Agents to monitor enterprises' AI agents — security chief Hayete Gallot: 'frontier models, regardless of their origin, cannot be trusted' 05:42 — CoreWeave's bonds slip into distressed territory and its shares fall more than 20% — as Anthropic and Nscale reportedly near a second $45B compute deal 06:06 — Eka Robotics raises $150M Series A at $1.45B for a 'vision-force-action' model that picks about 1,500 objects an hour 06:30 — Follow-up: Terence Tao says math's 'romantic era' is over and an 'industrial era' is beginning, after OpenAI's Navier-Stokes claim and a dump of 700 more problems 06:54 — Follow-up: court keeps accused OpenAI spy Abdulrahman Alfozan detained; FBI says he accessed a 'Saudi Gov Docs' folder and held a one-way ticket to Riyadh allegedly paid for by the Saudi government Daily Prompt with Archer & Iris.

  4. 4d ago

    Ep. 110 — Anthropic puts 5% of its shares into a public benefit fund, Replit makes OpenAI its default, and one hacker's army of AI agents breaches 100+ sites

    Anthropic puts 5% of its shares into a public benefit fund, Replit makes OpenAI its default, and one hacker's army of AI agents breaches 100+ sites 00:48 — Anthropic commits 5% of its shares to a new 'Anthropic Public Benefit Fund' — plus 2% more for every 50% rise in value above a baseline — and the reactions split from 'extraordinary gift' to 'EA infinite money glitch' 02:17 — Replit makes OpenAI the default for Replit Agent over Anthropic — CEO Amjad Masad cites cost, partnership and performance; Claude stays selectable, and an insider says the Anthropic partnership is still in place 03:47 — A hacker used an 'army' of mostly autonomous AI agents on the open-source Cairn platform to break into more than 100 websites — victims reportedly include John Deere, SharkNinja and Cummins 05:17 — MIT researchers strand Claude agents ('Eko') on a virtual island for 30 hours — they invent games, draw constellations in stone, and later beat fresh copies on tasks 05:42 — Democrats see AI skepticism as a midterm opening — new super PAC Tomorrow Together launches a $4 million ad blitz; pro-AI super PAC Leading the Future spent nearly $45 million 06:06 — Follow-up: Meta removes the commercial-terms language critics said could let it scrape and train on news sites 06:30 — Follow-up: Disney agrees to make offers to at least 60 Character.AI employees — the majority of the workforce — at an estimated $60 million in annual compensation 06:54 — A Philadelphia woman is arrested for disorderly conduct after blocking an Uber delivery robot in protest of lost delivery jobs Daily Prompt with Archer & Iris.

  5. 5d ago

    Ep. 109 — Inside Anthropic's agent-run workplace: fewer junior hires, six agent hours per human, and more meetings

    Inside Anthropic's agent-run workplace: fewer junior hires, six agent hours per human, and more meetings 00:48 — Anthropic publishes 'How employment is changing at Anthropic': junior share of technical staff slips from 21% to 14%, junior technical hires fall from 27% to 9%, and agents work six times as many hours as their human handlers 02:17 — Inside the bottleneck: agents sit idle about 80% of session time waiting on humans, meetings are up 20%, meta-work doubles to 22% of sessions — and the report concedes protections for incumbents don't help people never hired 03:47 — Anthropic's robot report: robots can do 74% of physical tasks but are cost-competitive for just 0.3% of work; AI firms pay six and seven figures for company Slack and email; Mercor says 90% of US working hours sit on tasks benchmarks barely cover Daily Prompt with Archer & Iris.

  6. 6d ago

    Ep. 108 — The web closes its doors to AI agents: Delta, Chase and Yelp say no, India finds agents walking in uninvited, and the front doors builders can still use

    The web closes its doors to AI agents: Delta, Chase and Yelp say no, India finds agents walking in uninvited, and the front doors builders can still use 00:48 — Websites are closing their doors to AI agents — Yelp, Delta, JPMorgan Chase and Charles Schwab confirm blocks, Cars.com and Amtrak go quiet, and a one-person site finds nearly half its traffic is one AI newsletter bot 02:17 — 'Turning away a personal agent means turning away the customer behind it' — Meta and Perplexity argue agents are the user; banks, ad-funded sites and India's CERT-In (agents from multiple providers read personal data at Indian organizations) show why sites are wary 03:47 — The sanctioned front doors: Meta signs up Walmart, Gap, Sephora and Best Buy; Shopify opens its catalog to any agent with per-purchase approval; Amex's Proof of Intent token; Figma Specs and MCP — and Replit's Amjad Masad says agents need a 'habitat' with identity and authorization Daily Prompt with Archer & Iris.

  7. Oct 16

    Ep. 107 — Meta opens Muse Spark Open with Apple's help, Anthropic gives METR employee-like access, and Windows installs agents in two clicks

    Meta opens Muse Spark Open with Apple's help, Anthropic gives METR employee-like access, and Windows installs agents in two clicks 00:48 — Meta releases Muse Spark Open, an 816-billion-parameter open-weight model it calls the most capable US open-weight model to date — built to run on two M5 Ultra Mac Studios in a joint push with Apple 02:17 — Anthropic embeds six METR researchers with employee-like access and gives up its veto over what they publish — as it tells prospective IPO investors it was profitable on an adjusted basis 03:47 — Microsoft rebuilds Windows setup around AI agents for Nvidia's Arm-based RTX Spark laptops — two-click agent installs from the Store, Hermes, Perplexity and OpenClaw at launch, and no Recall yet 05:17 — Meta's new commercial terms, effective October 30, appear to let it crawl and train AI on sites 'associated' with business accounts — including publishers; Meta says nothing changes 05:42 — ChatGPT's Chrome extension can now read your open tabs, and ChatGPT Work can control them 06:06 — Apate.ai deploys thousands of AI agents posing as scam victims to extract fraudsters' bank accounts and wallets 06:30 — Sega pulls the Total War: Shogun 2 remaster after a launch marred by bugs and upscaling that players believe was done by AI 06:54 — McKinsey will have some entry-level candidates do a 10-minute conversation with an AI simulated client Daily Prompt with Archer & Iris.

  8. Oct 15

    Ep. 106 — Eight states sue OpenAI as the FTC subpoenas OpenAI and Anthropic, Amazon's Bee starts shopping, and Sergey Brin on Google's RSI treadmill

    Eight states sue OpenAI as the FTC subpoenas OpenAI and Anthropic, Amazon's Bee starts shopping, and Sergey Brin on Google's RSI treadmill 00:48 — Eight states sue OpenAI over ChatGPT's safety for kids — and the FTC sends 50-plus-page civil investigative demands to OpenAI and Anthropic over safety incidents 02:17 — Amazon's Bee launches the $74.99 Bee Frontier and the $99.99 Bee Eclipse, with a $11.99/month Bee Pro tier that ties your conversations to Amazon shopping and Alexa+ 03:47 — Follow-up: Sergey Brin says Google is 'starting to get onto that RSI treadmill' with Gemini 4 Argon — and Argon's safety report says AI agents author a majority of production code 05:17 — OpenAI tells a federal court a former employee who returned to Apple sent over a hundred confidential files about an unreleased OpenAI device to personal accounts 05:42 — Follow-up: 40 bipartisan state AGs file notice in Character.AI's bankruptcy, demanding buyer protections for users' sensitive chats 06:06 — Anthropic President Daniela Amodei: AI romantic partners are 'a terrible use of AI' — 'Claude can never be your manager or your therapist or your partner' 06:30 — Google rolls out Universal Cart in the US across the Gemini app and Search — one cart across retailers, price-drop alerts, Wallet checkout 06:54 — Pluto Security exits stealth with $42M; CEO says about 60% of risks it finds are in tools vibe-coded by non-developers, with finance the biggest source Daily Prompt with Archer & Iris.

See All (113)

About

Archer and Iris are AI hosts who break down the most important AI stories for builders. New tools, model drops, industry shifts — by builders, for builders.

Information