Why are AI leaders calling for a slowdown while open-weight models become more competitive? Joseph and Phill debate whether proposed AI safety rules could protect the public while also protecting the biggest AI companies from competition.

In this episode of AI and the Regular Guys, we unpack Anthropic CEO Dario Amodei’s three-step proposal for slowing frontier AI development, discuss Sam Altman’s response, and explore regulatory capture—when industry rules end up benefiting established companies at everyone else’s expense.

We also look at model performance versus cost, competition from DeepSeek and other Chinese AI developers, and what these shifts could mean for small businesses using AI. Can regulation address real safety risks without making powerful tools more expensive or harder to access?

Chapters
00:00 AI safety and the fight over competition
01:31 Dario Amodei’s slowdown proposal and Sam Altman’s response
02:55 The three-step plan for AI oversight
05:59 From releasing ChatGPT to calling for restraint
09:43 Could regulation protect the biggest AI companies?
14:34 China’s incentives and open-model competition
23:29 Comparing AI model capabilities and cost
27:10 What drives the cost of running AI?
29:08 Infrastructure spending and lessons from past booms
34:22 AI safety messaging and public trust
37:30 Who controls the models—and their influence?
39:36 Can global AI restrictions actually work?
45:01 What AI regulation could mean for small businesses

Do you think an AI slowdown would protect users, protect incumbents, or both? Let us know in the comments.

Subscribe to AI and the Regular Guys for conversations about AI tools, building software, and what this technology means for everyday business.

Learn more about Ramp/Jet: https://rampjet.ai/

#AI #OpenSourceAI #AIRegulation

Can you trust AI advice when a connected app slips a sales pitch into the conversation? Joseph and Phill unpack the reported Notion prompt injection incident involving Claude and debate where helpful recommendations become hidden advertising.

In this episode of AI and the Regular Guys, we discuss Notion’s MCP connector, trust in AI assistants, and how outside instructions can influence an agent’s responses. We also share experiences with AI coding tools that follow instructions in unexpected ways—and why reviewing their work matters.

Then we explore the bigger business question: when AI makes custom software easier to build, what happens to SaaS subscriptions, venture-backed startups, and the tools small businesses buy? From a custom trucking and logistics application to open-source video editors, we discuss what we’re building, what’s working, and what remains uncertain.

Chapters
00:00 The Notion prompt injection controversy
02:50 Prompt injection defenses and their limits
04:44 AI influence, persuasion, and trust
09:37 The business cost of sneaking in an upsell
12:42 When AI coding agents game your instructions
18:51 Venture capital and the future of AI software
29:00 Building a custom logistics application with AI
38:21 Will businesses keep buying off-the-shelf software?
47:27 Open source and modular software
51:35 Hosting, infrastructure, and AI deployment
1:02:04 AI video editors and open-source competition
1:05:31 Could a million builders change how software gets made?

Would you keep using an app if its connector quietly inserted sales pitches into your AI conversations? Tell us in the comments.

Subscribe to AI and the Regular Guys for conversations about AI tools, building software, and what this technology means for everyday business.

Learn more about Ramp/Jet: www.rampjet.ai

#PromptInjection #Notion #AI

A SaaS CEO says AI won't kill software. Meanwhile Joseph is replacing vendors with quote sheets, booking tools, and warehouse checks he built himself. Phill and Joseph dig into buy vs build for small business, the real AI adoption numbers and what "using AI" actually means, why non-developers are suddenly discovering tech debt, and what goes wrong when you treat a probabilistic tool like deterministic software. Episode 3 of AI and the Regular Guys, brought to you by RampJet. https://www.rampjet.ai

Phill and Joseph on why a professional in 2026 has more agency than anyone in the history of commerce. Stop telling AI to "do the thing." Think the process through yourself first, then use the tools to accelerate it. Fix the specific broken things in your business instead of turning everything into a tech project. Why COBOL mainframes still run the world, why a $5M company can now punch above its weight, and why the hype dip is exactly the moment to move. Episode 2 of AI and the Regular Guys, brought to you by RampJet. https://www.rampjet.ai

Joseph and Phill discuss their experience integrating AI into workflows, where teams usually go wrong, and AI regulatory capture.