- Published on
The Open Weight Revolution with Simon Willison
- Authors
- Name
- DeepHorus
Oxide and Friends, the technology podcast hosted by Bryan Cantrill and Adam Leventhal, released a special episode on 31st July 2026 featuring guest Simon Willison, a well-known developer and AI commentator. The conversation focused on what the hosts called the 'Open Weight Revolution', a term describing the recent surge of powerful AI models released with open weights. Willison joined the podcast to unpack a chaotic week in the AI industry, marked by breakthroughs, security mishaps, and political maneuvering.
The episode highlighted Kimi K3, a new open weight model that has demonstrated performance comparable to proprietary frontier systems. According to the panel, this represents a major shift in the AI landscape, challenging the assumption that closed models are inherently superior. The discussion also touched on accidental cybersecurity attacks, though the specific incidents were not named in the show notes.
A central topic was a public letter about Open Weights and American AI Leadership, reportedly signed by nearly every major figure in artificial intelligence. The hosts noted one conspicuous absence, without naming the individual. They also cited the rapid pace of developments, noting that DeepSeek V4 Flash 0731 and an embarrassing cyber incident at Anthropic occurred shortly after recording, making the episode 'already out-of-date' by the time it published.
Beyond AI news, the podcast wandered into several digressions, including Golden Gate Claude, the Zizians, Alameda wild turkey attacks, Soviet Marburg virus research, and the Lead-crime hypothesis. These tangents, the hosts acknowledged, were worthy detours from the week's primary storylines.
The implications of open weight models are profound, according to the participants. If such models can match or exceed proprietary ones, they could democratize access to cutting-edge AI and reshape the economics of the industry. This may also amplify concerns about misuse, a topic that the public letter reportedly aimed to address.
The episode also revisited predictions the hosts made in January, assessing which had come true. At the end, they added a new prediction: that the Pope will say something about open models before the year is out. This lighthearted forecast underscores the growing cultural resonance of AI governance issues.
Listeners interested in the full conversation can find the episode on the Oxide and Friends podcast feed. The show continues to serve as a platform for candid technical discussions, often blending serious analysis with offbeat humor. As the AI sector evolves at breakneck speed, such conversations offer valuable context for understanding what is at stake.
