Episodes
-
Did OpenAI's Secret Model Just Crack Math?
OpenAI has produced solutions to over 100 long-standing math problems using an unreleased model, highlighting both advances and concerns about research ethics in AI-driven mathematics.
-
Le Chonk
Reflection's Beam is a major US open-weight AI model that rivals Chinese models in reasoning while using less compute, aimed at business and developer markets.
-
Why Did Another OpenAI Safety Insider Quit?
David Robinson argues that AI labs should adopt safety practices like nuclear power plants, with redundancy and thorough planning, because the trial-and-error approach is no longer sufficient.
-
Did OpenAI Fire Its Whistleblowers?
OpenAI dismissed three researchers for mishandling sensitive company information, raising concerns about internal security and transparency.
-
Is Gemini 4 Argon Just Benchmaxxing?
Google claims Gemini 4 Argon offers industry-leading performance, but internal and external reviewers are skeptical about its real-world coding and usability performance.
-
Dippin Dots
President Trump’s voluntary AI safety accord is a symbolic step that relies on the self-policing of companies rather than enforced regulation.
-
We Have The Anthropic IPO Numbers
Anthropic’s IPO reported a $42 billion net loss despite a 12x revenue increase, reflecting huge operational costs and heavy investment in AI infrastructure.
-
Starship Reaches Orbit For The First Time
Nvidia's Open Agent Safety Platform aims to contain AI agents, preventing them from escaping their sandbox environments, which is critical for safety and security.
-
Is Copilot Microsoft's Everything App?
Microsoft's co-pilot super app consolidates chat, coding, and agent functionalities into a single interface, integrating new features tailored to knowledge workers.
-
Meta Made a Tamagotchi for Its AI
Meta's new VR glasses aim to provide a lighter, more user-friendly alternative to traditional headsets with immersive displays and gesture controls.
-
AI Models Just Got Cheaper?
Anthropic's Claude Opus 5.5 offers equivalent performance to Fable 5.1 at a 40% lower cost and with enhanced safety safeguards.
-
Meta Finally Has An AI Hit
Meta’s Muse AI has seen rapid adoption, with over 902,000 downloads in six days, indicating strong user interest in highly personalized, social media-integrated AI agents.
-
The Googlebooks Are Here
Google's new Googlebook laptops are heavily focused on Android integration, targeting Android phone users with features like device syncing, app continuity, and notifications sharing.
-
OpenAI Got Hacked By... AI
AI security vulnerabilities are increasingly sophisticated, with researchers demonstrating how AI tools can be exploited to access sensitive corporate data.
-
Catfished By AI!
OpenAI is still grappling with significant issues related to AI alignment, and industry transparency is crucial for progress.
-
Crypto Takes An "L" In The Senate
The U.S. Senate’s failure to advance the Crypto Industries Clarity Act delays clearer regulatory frameworks for the crypto industry, likely causing industry uncertainty until next year.
-
Trump Says: No Slowdown!
AI safety debates are deeply divided, with industry leaders largely dismissing safety risks and regulatory advocates warning of potential existential threats.
-
The Great AI Slowdown?
There is a significant push within the AI industry to slow development for safety reasons, with industry leaders like Sam Altman, Elon Musk, and Dario Amadi advocating for global coordination and embedded safety evaluators.
-
It's A Doozy
Anthropic's threat report reveals sophisticated state-sponsored AI misuse campaigns, including cyberattacks, surveillance, and biological research efforts, highlighting urgent safety concerns.
-
The Man Who Quit The AI Industry
AI development is progressing rapidly, but significant concerns remain about safety, transparency, and the potential for catastrophic risks.
Follow this show on AudioBrevity and new episodes get summarized for you. Create a free account