Anthropic's $30B Ramp, Mythos Doomsday, OpenClaw Ankled, Iran War Ceasefire, Israel's Influence
All-In with Chamath, Jason, Sacks & Friedberg Podcast Recap
Published:
Duration: 1 hr 29 min
Guests: Brian Gerstner, Brad Gerstner
Summary
Anthropic's latest AI model, Mythos, is withheld due to its potential dangers, revealing vulnerabilities in major systems. The episode also covers geopolitical developments including a ceasefire in the Iran war and Israel's influence on US policy.
What Happened
Anthropic is withholding its latest AI model, Mythos, because of its ability to autonomously find thousands of vulnerabilities across major operating systems and web browsers. Mythos discovered long-standing vulnerabilities, such as a 27-year-old bug in OpenBSD and a 16-year-old flaw in FFMPEG, emphasizing the model's potential threat if released prematurely.
Anthropic has initiated Project Glasswing, a coalition with tech giants like Apple, Microsoft, and Google, to use AI for finding and fixing software vulnerabilities over a 100-day period. This move is part of Anthropic's strategy to sandbox AI capabilities and build defensive alliances, aiming to advance AI safely.
Despite Anthropic's proactive measures, there's skepticism around its fear-based marketing strategy. Chamath Palihapitiya suggests that the urgency surrounding Mythos might be exaggerated, comparing it to OpenAI's cautious release of GPT-2.
The episode touches on Anthropic's significant market growth, with a revenue run rate exceeding $30 billion and over 1,000 enterprises paying significant sums for their services. The discussion notes the potential for AI-generated code to dominate the market, increasing from 5% to 95% in the coming years.
OpenAI is also facing internal challenges, with staff departures and strategic uncertainties. However, its upcoming model, Spud, is anticipated to be a strong competitor to Mythos, suggesting a competitive landscape in AI development.
On the geopolitical front, a two-week ceasefire has been negotiated in the Iran war, with discussions held in Islamabad. Despite initial fears, markets have rebounded, reflecting confidence in the current US strategy.
The influence of Israel on US foreign policy is examined, with concerns about rising anti-Semitism linked to Netanyahu's actions. Former Israeli Prime Minister Naftali Bennett acknowledges Israel's declining popularity in the US, adding a layer of complexity to international relations.
Key Insights
- Anthropic's Mythos model has identified major vulnerabilities in critical systems, including a 27-year-old bug in OpenBSD, highlighting the potential risk of advanced AI models if released without caution.
- Project Glasswing is a collaborative effort involving major tech companies aiming to use AI to address software vulnerabilities over a short, focused period, emphasizing a new model for AI-driven cyber defense.
- The market for AI-generated code is projected to grow dramatically, with estimates suggesting it could account for up to 95% of the total code market, fueled by Anthropic's current dominance and rapid growth.
- A two-week ceasefire in the Iran war has been achieved, with markets responding positively, suggesting a level of confidence in the diplomatic strategies employed, despite geopolitical complexities.
View all All-In with Chamath, Jason, Sacks & Friedberg recaps