Personalized GPTs Are Here + F.T.C. Chair Lina Khan on A.I. Competition + Mayhem at Apefest

Episode Summary

Executive Summary: This episode centers on OpenAI’s first Dev Day and the launch of custom GPTs, which promise personalized, task-specific chatbots using private documents and limited actions. The hosts discuss benefits, startup disruption, safety concerns about more agentic AI, and job displacement. In a second segment, FTC Chair Lina Khan explains how regulators are approaching AI competition, privacy, copyright, and consumer protection. The episode closes with a surreal crypto story about Bored Ape event attendees suffering eye injuries from UVA lighting, underscoring broader industry recklessness.

Main Topics: OpenAI Dev Day and product announcements (Priority: 5/5): The hosts recap OpenAI’s first developer conference, noting the scale of the event and incremental updates like GPT-4 Turbo, pricing changes, and Copyright Shield. The tone is that OpenAI has become a major platform company whose product moves matter to the whole ecosystem. Custom GPTs as the next AI interface (Priority: 5/5): A major focus is OpenAI’s new custom GPTs: personalized chatbots that can use uploaded private documents, follow custom instructions, and eventually potentially take actions. The hosts demo examples like a daycare handbook bot, an AI grandfather for investment advice, and a Hard Fork podcast bot. Safety, agency, and future AI risks (Priority: 5/5): The discussion shifts from today’s limited GPT features to future AI agents that could plan and act on users’ behalf. The hosts and Lina Khan both emphasize current limitations but warn that more powerful agents could expand attack surfaces for scams, manipulation, and other harms. Startup disruption and platform power (Priority: 4/5): OpenAI’s move to build features directly into ChatGPT is framed as both helpful for users and damaging to startups that built wrapper products. The conversation highlights tension between innovation, consolidation, and the emerging platform/app-store model for AI. FTC Chair Lina Khan on AI regulation (Priority: 5/5): Lina Khan argues the FTC should use existing competition, privacy, and consumer-protection tools to police AI harms now, rather than waiting for new legislation. She focuses on fraud, discrimination, privacy, and market concentration as immediate issues. Crypto’s continuing dysfunction (Priority: 3/5): The closing segment covers ApeFest attendees reportedly suffering eye injuries from UVA-emitting lights, followed by discussion of Sam Bankman-Fried’s fraud conviction. The takeaway is that crypto remains chaotic, risky, and poorly executed even after the hype has faded.

Key Arguments: OpenAI’s scale is now so large that its developer conferences are system-shaping events, not just product demos. Custom GPTs are a meaningful step forward because they let users personalize chatbots with private documents and tailored instructions, even if the technology is still limited. These GPTs are more evolutionary than revolutionary; they often feel like GPT-4 with a bit of personalization rather than a fully new capability. The future risk is not just chatbots that answer questions, but agents that can take actions, which could be abused for scams, coercion, or other harmful behavior. AI will likely displace or compress jobs that involve repeated Q&A against standardized documents, such as customer support or benefits administration. Lina Khan argues regulators should focus on harms already occurring—voice-cloning scams, discrimination, privacy violations, and anti-competitive behavior—rather than speculative apocalypse scenarios. The FTC sees openness and open source as potentially beneficial for competition, but warns that “open” can be used as a veneer for new forms of concentration. The crypto segment suggests the industry has failed to build trust, safety, and useful products, even after years of pressure to do so. OpenAI’s new app-store-like GPT ecosystem could create a new gatekeeper problem similar to the last platform era. Existing laws can and should be used upstream to prevent harm, rather than relying solely on after-the-fact punishment.

Data Points: OpenAI weekly users: 100 million - Sam Altman said ChatGPT now has 100 million weekly users at Dev Day. Fortune 500 adoption: 92% - Altman said 92% of Fortune 500 companies are building on OpenAI’s products. Dev Day YouTube views: over 700,000 - The hosts noted the keynote had over 700,000 views on YouTube in one day. GPT rollout access: ChatGPT Plus and Enterprise only - Custom GPTs were described as unavailable to free users. OpenAI model update: GPT-4 Turbo - Announced as a more up-to-date, larger-context, cheaper model. P-doom estimate: 15% - Lina Khan said her probability that AI will kill us all is about 15%. ApeFest injuries: UVA-emitting lights in one corner of the event - Yuga Labs said these lights were likely the cause of eye and skin issues. Bored Ape purchase example: around $100,000 - A Bored Ape owner interviewed by Kevin said he spent roughly this amount on his NFT. Lina Khan’s FTC tenure: youngest FTC chair in history - The hosts described her as the youngest chair in the agency’s history.

Pivotal Quotes: "We are witnessing the formation of a generationally important company." — Casey Newton: Used during the discussion of OpenAI’s scale and the significance of Dev Day. "We’re just not at a point where we want these things to do any kind of high-level planning." — Sam Altman: Altman’s answer about limiting GPT agent capabilities to safer, simpler actions. "There’s no AI exemption from the laws already on the books." — Lina Khan: Her explanation of how the FTC plans to enforce existing rules against AI-related harms.

Implications: AI is moving from novelty to infrastructure, with personalized assistants and agent-like tools likely to reshape software markets, labor, and regulation. The next phase will hinge on whether competition and safety keep pace with capability.