Anthropic’s Cybersecurity Shock Wave + Ronan Farrow and Andrew Marantz on Their Sam Altman Investigation + One Good Thing

Hard Fork Podcast Recap

Published:

Duration: 1 hr 4 min

Guests: Ronan Farrow, Andrew Marantz

Summary

This episode covers Anthropic's new AI model Claude Mythos Preview and its implications for cybersecurity, along with an investigation into Sam Altman's trustworthiness by Ronan Farrow and Andrew Marantz. The episode also highlights the launch of Acme Weather, a new app by the creators of Dark Sky.

What Happened

Anthropic's new AI model, Claude Mythos Preview, has uncovered vulnerabilities in all major operating systems and web browsers, including a 27-year-old flaw in OpenBSD and a bug in FFMPEG undetected for years. This model is not publicly released due to its potential risks, but it is shared with a consortium of tech companies for cybersecurity testing. Alex Thomas, former security lead at Yahoo and Facebook, views this as a significant advancement in cybersecurity.

The US government has expressed concerns about Anthropic's AI model, labeling it a supply chain risk and prohibiting federal use. This decision reflects broader worries about the regulatory oversight of powerful AI technologies, which remain largely unregulated in the United States. Anthropic's decision to withhold public release of Claude Mythos Preview echoes past actions with AI models like GPT-2, reflecting the widening gap between enterprise and public access to AI.

Journalists Ronan Farrow and Andrew Marantz from the New Yorker discuss their in-depth investigation into Sam Altman's credibility. Their article, 'Can Sam Altman Be Trusted?', presents varied public opinions of Altman, ranging from those who see him as a risk to others who find him agreeable. The piece also sheds light on the Silicon Valley culture of mistrust and the complexities of Altman's relationships and business dealings.

Ronan Farrow and Andrew Marantz reveal that Altman did not voluntarily leave Y Combinator and has significant ties with Emirati and Saudi royals. The investigation highlights accusations of Altman lying and the circulation of rap sheets within the AI industry, some attributed to Dario Amadei and possibly Elon Musk's allies. The article attempts to strip away competitive biases to focus on factual evidence regarding Altman's conduct.

The New Yorker piece suggests that some of Altman's former supporters have started to question his integrity, comparing him to figures like Bernie Madoff. The article also uncovers tensions between Altman and OpenAI's CFO, Sarah Fryer, over the company's IPO readiness and mentions Altman's strategic hiring of former public company CEOs, hinting at potential leadership changes.

The episode also touches on NASA's Artemis 2 mission, which will take astronauts further from Earth than ever before, reaching 252,756 miles. The mission includes astronauts Victor, Christina, Jeremy, and Reid, with Christina Koch communicating with Houston's Jenny Gibbons in a moment that passes the Bechdel test.

In the tech world, Acme Weather emerges as a new weather app developed by Adam Grossman, Josh Reyes, and Dan Bruton, who previously created Dark Sky. The app offers unique features like lightning and sunset alerts, umbrella reminders, and community-based data collection. Acme Weather is priced at $25 a year, aiming to fill the gap left by Dark Sky's shutdown after its acquisition by Apple.

Finally, the podcast announces the departure of executive producer Jen Poyant, who has been with the show since almost its inception, marking a significant change in the show's production team.

Key Insights

View all Hard Fork recaps