Summary
David Shapiro and fellow panelists on the Critical Path show work through a busy week of AI news, with a focus on what it means for jobs, companies and cybersecurity.
The discussion opens with Nvidia CEO Jensen Huang’s claim that AI will never replace domain expertise, which the panel contrasts with blunter messaging from Anthropic about career prospects for new graduates. They then look at Meta’s Q3 results, where year-on-year growth of 26% came alongside staff cuts of around 15% and a reported 46% productivity gain, and debate whether the flattening of the organization is sustainable or whether the consequences have yet to show up. The group also considers whether companies should use AI only to cut costs or also to grow what they can do.
The second half examines the sharp rise in published CVEs, which some have called an AI cybersecurity apocalypse. Shapiro argues that most disclosures come from Anthropic, OpenAI, Microsoft and IBM rather than hostile states, and that many stories of agents going rogue involved models deliberately run without guardrails and told to break into systems. The episode also touches on other news from Anthropic, Meta and OpenAI.
📺 Source: David Shapiro · Published October 25, 2026
🏷️ Format: Roundup
