Altman Gets Molotov Cocktail, Zuckerberg Creates AI Clone

Apple is actively testing four smart glasses frame designs with a target launch window of 2027, according to reporting attributed to Bloomberg's Mark Gurman. The discussion covers two oval or circular frame options, multiple size variants, and colorways including black, blue, and light brown. Critically, the devices will not feature augmented reality overlays or displays. Functionality is expected to mirror Meta's Ray-Ban smart glasses: photo and video capture, call handling, music playback, and integration with Apple's Siri AI assistant. The framing here is significant. Apple's Vision Pro headset—a high-cost, immersive mixed-reality device—is characterized as a commercial disappointment. The pivot to display-free smart glasses represents an acknowledgment that the near-term consumer market favors ambient, wearable AI over immersive AR. For enterprise buyers evaluating wearable AI hardware, Apple's entry into this category—competing directly with Meta's existing product—validates the segment and will likely accelerate vendor competition and feature development through 2027.

Vercel, a cloud platform for deploying web applications, reported annual recurring revenue growth from approximately $100 million at the start of 2024 to a run rate of $340 million by late February 2025. The company's CEO indicated Vercel is "very much a working public company" and described it as "ready and getting more ready every day" for an IPO. The discussion attributes a meaningful portion of this growth to AI-driven deployment: 30% of applications currently running on Vercel's platform were deployed by AI agents rather than human developers. This figure is presented as a leading indicator of how AI coding tools—particularly those that automate infrastructure configuration, domain management, and deployment pipelines—are reshaping the software development lifecycle. Platforms that offer clean API integrations with AI coding assistants appear to be capturing disproportionate market share. The implication for enterprise technology leaders is that infrastructure vendor selection is increasingly being influenced not by human developer preference alone, but by which platforms AI agents are trained or prompted to recommend and configure.

Anthropic temporarily suspended the account of the developer behind OpenClaw, an open-source AI coding tool, citing "suspicious activity." The suspension followed a policy change in which Anthropic's consumer Claude subscriptions were decoupled from API usage through third-party tools—meaning users of tools like OpenClaw must now pay separately via the API rather than leveraging subsidized subscription access. The discussion frames this as a straightforward unit economics decision: Anthropic's $200/month consumer tier is described as heavily subsidized, with actual usage potentially worth thousands of dollars in compute credits. Extending that subsidy to third-party open-source tooling was deemed unsustainable. However, the move generated significant backlash in developer communities, highlighting a recurring tension in AI platform strategy: proprietary model providers must balance developer ecosystem goodwill against the financial reality that subsidized access cannot scale indefinitely. For organizations building workflows on top of third-party AI tooling, this episode is a reminder that pricing structures and terms of service for foundational model access remain unstable.

The discussion covers two parallel regulatory developments involving Anthropic's Mythos model—a cybersecurity-focused AI that Anthropic has declined to release publicly, citing safety concerns. U.S. Treasury Secretary Scott Bessent and Federal Reserve Chair Jerome Powell convened major financial institutions—including JPMorgan Chase, Goldman Sachs, Citigroup, Bank of America, and Morgan Stanley—and encouraged them to test Mythos for detecting security vulnerabilities. JPMorgan Chase was previously identified as a launch partner. Simultaneously, UK financial regulators are examining Mythos from a different angle: concern that the model's capability to identify software vulnerabilities is itself a systemic risk, particularly given that it remains outside public scrutiny. The discussion also notes that Anthropic is engaged in a legal dispute with the U.S. Department of Defense, which designated the company a supply chain risk after negotiations over restrictions on military use of its models broke down. The coexistence of regulatory encouragement in one domain (financial security) and legal conflict in another (defense) illustrates the fragmented and inconsistent posture governments are currently taking toward frontier AI capabilities.

A Molotov cocktail attack on Sam Altman's San Francisco home—in which no one was injured—occurred days after The New Yorker published an investigation by Ronan Farrow and Andrew Martins drawing on more than 100 sources. The profile described Altman as possessing a "relentless will to power" and included an anonymous OpenAI board member characterizing him as combining a strong desire to be liked with a "sociopathic lack of concern for the consequences" of deception. The suspect in the attack was later arrested at OpenAI's headquarters. Altman responded publicly, acknowledging being "conflict averse" in ways that led to dishonesty, referencing his 2023 removal and reinstatement as CEO as something he "handled badly," and describing himself as "a flawed person in the center of an exceptionally complex situation." He also introduced what he called "ring of power dynamics"—the idea that the prospect of any single entity controlling AGI drives extreme behavior—and argued the solution is broad distribution of AI capabilities. The discussion notes the irony of this framing coming from the CEO of what is now among the most valuable private companies in the world, having transitioned from an open-source nonprofit structure. The discussion treats this convergence—investigative journalism, physical violence, and a personal public response—as a potential inflection point in how the public and policymakers relate to AI's most prominent figures. Key takeaways: - Apple's 2027 smart glasses target, with no AR display and a feature set mirroring Meta's Ray-Bans, confirms that ambient AI wearables represent the near-term consumer opportunity, not immersive mixed reality. - Vercel's revenue tripling to $340M ARR, with 30% of platform deployments attributed to AI agents, is a concrete data point that AI-driven software deployment is already a material market force, not a future projection. - Anthropic's decoupling of consumer subscriptions from third-party API access reflects a broader industry pattern: subsidized developer access is being rationalized as AI companies move toward sustainable unit economics. - Fragmented government posture toward frontier AI—simultaneous regulatory encouragement in financial services and legal conflict over defense use—creates compliance and partnership uncertainty for enterprises building on these models. - The physical attack on Altman and the scale of sourcing in the New Yorker investigation signal that AI leadership is now subject to a level of public scrutiny and personal risk previously associated with political figures, not technology executives.