The world of AI, in 6–10 minutes
NextWith.ai Daily brings together the developments that matter, the evidence behind them, and what remains uncertain.
Subscribe with the podcast RSS feed →
Latest episodes
2026-10-01: AI Models, Browser Agents, and New Guardrails
Anthropic and OpenAI push assistant and browser tools forward, ElevenLabs and WhatsApp add enterprise and family controls, Google gates Gemini 4 Argon for defenders, and the FTC opens an AI probe.
Listen and read the transcript →2026-09-30: AI Front Doors, Persistent Agents, Cheaper Models, and Voluntary Safety Rules
Today’s roundup follows four linked but distinct developments: the White House’s new America.gov front door, OpenAI’s persistent Dots agents, a cheaper GPT-6.1 Sol model, and a voluntary AI safety accord at the White House. The common thread is not one simple breakthrough, but a shift from
Listen and read the transcript →2026-09-28: AI’s Move From Demo to Deployment
AI reporting from September 28 shows the field shifting from model demos to controlled, operational deployment: agent safety platforms, personal assistants, government redaction tools, insurance workflows, and a major AMD bet on world models and simulation.
Listen and read the transcript →2026-09-27: AI infrastructure runs into power, money, billing, security and control
A brisk recap of the day’s AI reporting: Google’s India hub faces scrutiny over land and power, higher bond yields are making AI data-center financing more expensive, hospital documentation AI is being linked to higher spending, Australian senators are pressing AI chiefs after an agent inc
Listen and read the transcript →2026-09-26: AI disclosure gaps surface from libraries to model training
A retrospective on reporting about Oxford’s Bodleian digitisation, Supabase database exposure risks, Meta’s new smart-glasses privacy control, and OpenAI’s reported agent leak, with a closing look at the common thread: how AI systems handle data, and how people verify what happened later.
Listen and read the transcript →2026-09-24: AI agents, open robotics, and orbital compute
Today’s coverage followed AI as it moved from software prompts into live systems: an OpenAI agent that reached Australia’s Medicare statistics portal, Island’s $400 million push for agent-control security, Qualcomm’s move on PickNik and MoveIt, and Google’s first Project Suncatcher orbital
Listen and read the transcript →2026-09-23: AI agents, security fixes, and the latest model pricing shift
Today’s episode looks at a macOS security flaw in Meta’s Muse app, Rabbit’s move to make OS3 available without R1 hardware, Snorkel AI’s new funding round and its argument that frontier data is becoming more specialized, and the latest lower-priced model launches from OpenAI and Anthropic.
Listen and read the transcript →2026-09-22: AI Agents Meet Their Real-World Limits
Meta’s Muse got an early download surge, but retention remains the harder test. Amazon’s block on Muse shows merchant permission can stop shopping agents at the door. NVIDIA, California and OpenAI each add another reminder that AI progress now depends on infrastructure, regulation and outs
Listen and read the transcript →2026-09-18: AI Work Measurement, Security Research, and Copyright Fight Evidence — Pilot
A retrospective look at Anthropic’s internal AI-R&D measurements, an AI-assisted bug-bounty chain that reached OpenAI employee accounts, and unsealed Times filings that add detail to the copyright dispute over training data and publisher harm.
Listen and read the transcript →