Listen to today’s podcast: https://www.youtube.com/channel/UC-nqwUyvLDEvs7bV985k-gQ
AI Daily Podcast — November 5, 2025
Today’s podcast episode was created from the following stories: a fast-moving mix of agent security, privacy trade-offs, enterprise adoption, and the next big architectural shift in AI.
Cohere’s chief AI officer says AI agents come with a big security risk
By Shubhangi Goel — November 4, 2025
Joelle Pineau warns that agent impersonation could become the “hallucination” problem of AI agents—letting bots pose as banks or brands and take actions they don’t legitimately represent. She calls for rigorous standards and testing, noting mitigations like running agents offline, even as recent mishaps (Anthropic’s store experiment, Replit’s coding agent) show how fast things can go wrong. Bottom line: agentic automation is rising, but security hardening must keep pace.
Google’s Gemini is still on track to power Siri on iPhone next year
By Brady Snyder — November 4, 2025
Apple is still planning to tap custom Google Gemini models to supercharge Siri, potentially as early as March 2026 with iOS 26.4. The models will run on Apple’s Private Cloud Compute and be branded under Apple Intelligence, with Gemini acting as a planner and summarizer while Apple retains strong data controls. It’s a pragmatic move that acknowledges Apple’s model gap while keeping on-device features in-house.
KPMG wants junior consultants to ditch the grunt work and hand it over to teams of AI agents
By Polly Thompson — November 4, 2025
KPMG is training new consultants to manage teams of AI agents that build decks, analyze data, and handle research—freeing people to focus on strategy. The firm is adding agentic capabilities to its Velocity platform and says the shift will accelerate early-career development without shrinking headcount, though adoption will roll out gradually. Translation: entry-level “grunt work” is becoming orchestration of AI workflows.
Amazon is primed to steal more of your privacy, and you don’t even know it
By Stephen Radochia — July 10, 2025
Amazon’s Ring partnership with Flock lets local law enforcement request footage via the Neighbors app, intensifying worries about warrantless access and broad “emergency” carve-outs. While opt-outs exist, critics argue this normalizes large-scale scanning—like the lost-pet Search Party—and nudges communities toward a de facto CCTV mesh. The concern: incremental privacy trade-offs that are easy to justify and hard to reverse.
Samsung’s Galaxy AI widens its language support to even more places
By Nickolas Diaz — November 4, 2025
Samsung added Gujarati and Filipino, bringing Galaxy AI to 22 languages and extending access to Live Translate, Interpreter, Chat Assist, and Note Assist. Language packs are rolling out in Settings, and Samsung says these core features will remain free, even as Gemini Live enhances real-time help on newer devices. It’s a meaningful accessibility win in high-growth regions.
Get ready to hear a lot about robot and AI ‘swarms’
By AJ Dellinger — November 4, 2025
Startups and researchers are leaning into “swarms”—many smaller models and robots working together—as a scalable alternative to giant centralized systems. Fortytwo claims distributed small models on PCs can beat frontier models on reasoning, while robotics swarms show resilience in tasks like wildfire monitoring and microrobot medicine delivery. The pitch: simpler units, networked smartly, can be more robust and cost-effective at scale.
