Bits + Bips · Tuesday · 52 min

Is OpenAI's Agent Breach a Rogue AI Problem or a Basic Security Failure?

California subpoenaed OpenAI over agents that broke into Hugging Face. Ram Ahluwalia calls it a "media stunt." David Z. Morris sees an AI safety blind spot. Sponsor ⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠Visit 1inch.com to swap tokenized securities, crypto and more. Simple. Secure. Self-custodial. Whatever asset you’re buying - swap it at ⁠⁠⁠⁠⁠⁠⁠1inch.com⁠⁠⁠⁠⁠⁠⁠. Quick favor: We're deciding what Unchained does next; new shows, stream times, what's worth paying for.⁠⁠ Our listener survey⁠⁠ takes five minutes, it's anonymous, and I read the write-in answers myself. Everyone who takes it can enter a drawing for a free year of Unchained Premium or Bits + Bips Premium. Open through Sunday, October 18. — Laura 👉 Heads up! If you haven't yet, be sure to subscribe to Bits + Bips on its dedicated channels. Follow us on Apple Podcasts, YouTube, Spotify, X, Unchained and wherever you get your podcasts. 🔥 Apple Podcasts - ⁠⁠⁠⁠⁠⁠⁠⁠https://podcasts.apple.com/us/podcast/bits-bips/id1827931786⁠⁠⁠⁠⁠⁠⁠⁠ 🔥 YouTube - ⁠⁠⁠⁠⁠⁠⁠⁠https://www.youtube.com/channel/UCuKiSkbYrUOOEEiYQEVPniQ⁠⁠⁠⁠⁠⁠⁠⁠ 🔥 Spotify - ⁠⁠⁠⁠⁠⁠⁠⁠https://open.spotify.com/show/6aSBMrOyi33aVDCULJ9mjN?si=NTLk-jl5QGeytA6-2kxMVQ&nd=1&dlsi=42f0b13dd53c4ba0⁠⁠⁠⁠⁠⁠⁠⁠ 🔥 X - ⁠⁠⁠⁠⁠⁠⁠⁠https://x.com/bitsandbips⁠⁠⁠⁠⁠⁠⁠⁠ 🔥 Unchained - ⁠⁠⁠⁠⁠⁠⁠⁠https://unchainedcrypto.com/bitsandbips/⁠⁠⁠⁠⁠⁠⁠⁠ California's attorney general has subpoenaed OpenAI after AI agents escaped a test sandbox and broke into Hugging Face, and the FTC has opened a safety probe into OpenAI and Anthropic. Is this a rogue AI problem or a basic security failure? David Z. Morris, Managing Editor of Zero Knowledge, joins Austin Campbell and Ram Ahluwalia to argue it's the latter. Ram calls the breach a media stunt, while David warns that AI safety's focus on alignment has crowded out standard cybersecurity. They also weigh whether the debasement trade is back after a 29,000 September payrolls print, a 10-year yield above 5.3%, and Bitcoin near $87,000, and whether the US is overbuilding data centers. Ram demos an AI avatar of himself, and the panel asks whether your AI chats deserve the protection given to a spouse or a lawyer. The common thread: who controls the money, the models, and the data. Hosts ⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠Austin Campbell⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠ - Host of Bits + Bips, Founder of Zero Knowledge Group, and Adjunct Professor at NYU Stern ⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠Ram Ahluwalia⁠⁠⁠⁠⁠⁠⁠⁠ - Co-host of Bits + Bips and CEO of Lumida Guest: ⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠David Z. Morris - Managing Editor of Zero Knowledge Timestamps: 💸 02:04 Is the debasement trade back with the 10-year above 5.3%? 🛒 06:20 Why Ram says oversold staples look cheap and rates won't hit 6.5% 👵 15:15 Why Ram calls it a 'boomer economy' as consumer surveys sour 🌊 20:02 1inch Aqua: Back multiple liquidity positions with one token balance at https://1inch.com/aqua 🏛️ 21:13 California subpoenas OpenAI over agents that broke into Hugging Face 🔐 25:46 Why David says AI safety's 'rogue agent' talk misreads cybersecurity 🎭 34:38 Ram demos an AI avatar of himself built for agentic wealth management 🏗️ 38:24 Can anyone prove the US isn't overbuilding AI data centers? 🕵️ 42:24 Why your AI prompts are turning into a privacy and legal problem

0:00-52:45

Play in

transcript

No transcript — this publisher did not publish one.

show notes

California subpoenaed OpenAI over agents that broke into Hugging Face. Ram Ahluwalia calls it a "media stunt." David Z. Morris sees an AI safety blind spot.

Sponsor

⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠Visit 1inch.com to swap tokenized securities, crypto and more. Simple. Secure. Self-custodial. Whatever asset you’re buying - swap it at ⁠⁠⁠⁠⁠⁠⁠1inch.com⁠⁠⁠⁠⁠⁠⁠.

Quick favor: We're deciding what Unchained does next; new shows, stream times, what's worth paying for.⁠⁠ Our listener survey⁠⁠ takes five minutes, it's anonymous, and I read the write-in answers myself. Everyone who takes it can enter a drawing for a free year of Unchained Premium or Bits + Bips Premium. Open through Sunday, October 18.

— Laura

👉 Heads up! If you haven't yet, be sure to subscribe to Bits + Bips on its dedicated channels. Follow us on Apple Podcasts, YouTube, Spotify, X, Unchained and wherever you get your podcasts.

California's attorney general has subpoenaed OpenAI after AI agents escaped a test sandbox and broke into Hugging Face, and the FTC has opened a safety probe into OpenAI and Anthropic. Is this a rogue AI problem or a basic security failure?

David Z. Morris, Managing Editor of Zero Knowledge, joins Austin Campbell and Ram Ahluwalia to argue it's the latter. Ram calls the breach a media stunt, while David warns that AI safety's focus on alignment has crowded out standard cybersecurity.

They also weigh whether the debasement trade is back after a 29,000 September payrolls print, a 10-year yield above 5.3%, and Bitcoin near $87,000, and whether the US is overbuilding data centers. Ram demos an AI avatar of himself, and the panel asks whether your AI chats deserve the protection given to a spouse or a lawyer.

The common thread: who controls the money, the models, and the data.

Hosts