Everyone Wants To Give Back

The episode begins with a lively introduction from host Brian McCullough, highlighting several major news items in the tech world for October 19, 2026. The host quickly summarizes Elon Musk's announcement about giving every American child a SpaceX share valued at around $10.5 billion, and notes that Anthropic has committed 5% of its shares to a public benefit fund. The episode promises coverage of diverse topics including MapQuest adding surveillance features, and significant AI developments from Google DeepMind and Anthropic.

Elon Musk stated he intends to give one of his SpaceX shares to every kid in America in 2027. Based on SpaceX's closing price of about $146.03, the total gift would be roughly $10.5 billion, considering the approximately 72 million children under 18 in the U.S. As of August, Musk owns about $4.8 billion worth of SpaceX shares, which could rise to $6.4 billion with unvested awards. Musk posted this plan on X (formerly Twitter) without specific details on execution, and this announcement coincided with SpaceX's stock rising on Friday. Additionally, Musk's announcement is linked to ongoing activities where SpaceX's president Gwen Shotwell has mentioned that some SpaceX stock has been given to Trump accounts benefiting millions of U.S. children.

Anthropic revealed it plans to commit 5% of its shares, increasing to roughly 20% with co-founder and staff pledges, to a public benefit fund. This fund may expand as the company's valuation grows, with an additional 2% allocated for every 50% increase in valuation. The fund will be structured as a wholly owned LLC rather than a traditional nonprofit, and will avoid lobbying or political donations. Notably, Anthropic’s long-term benefit trust, which can influence company decisions and control four out of seven board seats, is governed by undisclosed documents. Trustees include prominent figures like Ben Bernanke, and the trust's mandate is to steer AI for humanity's benefit while maintaining disinterest in profits. Shareholders might be able to modify the trust with supermajorities, but the details are unspecified, rendering the trust’s influence complex and somewhat opaque.

MapQuest has added a new surveillance feature to its mobile app that displays locations of license plate readers and cameras from over ten companies, including Flock, Vigilant, and Axon. These ALPRs are shown as red eye icons on the maps, with Flock operating over 120,000 cameras across the U.S. This update aims to improve crime-fighting capabilities by providing more information on where these cameras are located. However, privacy advocates express concern over potential misuse, such as tracking ex-partners or monitoring the public without consent. The move has received mostly positive reactions online, with some users saying it might even help win back users from Google, while others prefer to avoid such invasive mapping.

Four major European telecoms—Deutsche Telekom, Orange, Telefónica, and Vodafone—are exploring a joint bid to launch satellite mobile services aimed at challenging Starlink’s dominance. They plan to acquire spectrum licenses and potentially partner with an existing satellite provider, not by building a new constellation. This initiative comes amidst heightened geopolitical tensions, with the US threatening retaliation if European governments favor European companies over U.S. firms like SpaceX. The European Union is also considering reserved spectrum bands for direct-to-device satellite connectivity, which could impact U.S. companies operating in this space.

Google DeepMind’s latest AI, Gemini, successfully defeated Minecraft’s Ender Dragon without human intervention, accomplishing this in approximately 27 hours and 14 minutes at a cost of $1,620 predicted for 2027. Four AI agents, working in coordination and sharing tips, executed complex tasks by splitting roles—such as gathering supplies or building fortresses—and communicated via in-game chat. They exchanged nearly 230 messages, died over 600 times, and wrote 2,783 programs during the process. An independent 'observer' agent monitored for fairness and documented the event. This achievement demonstrates the advancement of general agentic AI models capable of complex autonomous tasks in challenging environments.

Anthropic’s AI model Claude cracked a long-standing crypto mystery involving the seeds used in NIST’s elliptic curve digital signature algorithm (ECDSA). The seeds, believed to originate from a random number generator seeded by a computer’s clock in the late 1990s, were thought to potentially contain backdoors. However, Claude’s analysis identified that the seeds resulted from a basic pseudorandom number generator based on timestamps from the late 90s, not a maliciously seeded backdoor. This finding suggests that, despite concerns over NSA involvement, the seeds were not rigged, and the cryptographic system remains secure. The discovery highlights AI’s ability to analyze complex, old data and solve cryptographic puzzles.