“Claude Opus 4.8: The System Card” by Zvi

“Claude Opus 4.8: The System Card” by Zvi

Only six weeks after Opus 4.7, we have Opus 4.8. For everyone, that means another incremental upgrade to Claude. It is once again smarter, and can do tasks for longer, and comes with a number of hot new features. For me, that also means reading another 244 page system card. It was only April 20 when I did a full review of the Opus 4.7 system card, plus an additional post focusing on related issues of model welfare. These updates are incremental and coming more rapidly, and this still is below the capability level of Claude Mythos, so the focus will be on the delta. What is different about Opus 4.8 versus what we already know about Opus 4.7 and Mythos? It turns out there's still a lot to talk about. Image created as self-portrait for this post by Claude Opus 4.8 Table of Contents 1. Here We Go Again: Executive Summary. 2. Introduction (1). 3. RSP Evaluations (2). 4. Move That Goalpost. 5. The Failures Are News. 6. Alignment Risk Slowly Rises. 7. New Risk Pathways Just Dropped. 8. Cyber (3). 9. Harmful Requests (4.1). 10. We Need To Talk (4.2 [...] --- Outline: (01:16) Here We Go Again: Executive Summary (02:33) Introduction (1) (02:42) RSP Evaluations (2) (03:47) Move That Goalpost (05:41) The Failures Are News (07:33) Alignment Risk Slowly Rises (09:00) New Risk Pathways Just Dropped (11:26) Cyber (3) (12:22) Harmful Requests (4.1) (14:23) We Need To Talk (4.2 and 4.3) (17:36) Overcoming Bias (4.4) (19:33) Agentic Safety (5) (21:40) Prompt Injection (5.2) (25:18) Alignment (6) (26:33) Looking For Problems (27:55) Who Watches The Training (6.2.2) (32:07) Automated Behavioral Audit (32:47) The Model Is Smarter Than The Eval (6.2.3.2) (34:39) You Should See The Other Guy (36:30) UK AISI Testing (6.2.4) (36:50) In Vendbench (6.2.5) (39:27) Honesty (6.3.3 to 6.3.6) (41:35) Chain of Thought (CoT) Monitorability (6.5) (44:09) What's In The Box? (6.6) (45:57) That's All For Now --- First published: May 29th, 2026 Source: https://www.lesswrong.com/posts/Gx6cJ6cG9JfeSNcLB/claude-opus-4-8-the-system-card [https://www.lesswrong.com/posts/Gx6cJ6cG9JfeSNcLB/claude-opus-4-8-the-system-card?utm_source=TYPE_III_AUDIO&utm_medium=Podcast&utm_content=Source+URL+in+episode+description&utm_campaign=ai_narration] --- Narrated by TYPE III AUDIO [https://type3.audio/?utm_source=TYPE_III_AUDIO&utm_medium=Podcast&utm_content=Narrated+by+TYPE+III+AUDIO&utm_term=lesswrong&utm_campaign=ai_narration]. --- Images from the article: Man holding gear before massive suspended clockwork mechanism with multiple rings and moon phase display. [https://res.cloudinary.com/lesswrong-2-0/image/upload/f_auto,q_auto/v1/mirroredImages/Gx6cJ6cG9JfeSNcLB/zzlyhakfrla31qrmwg9c]https://res.cloudinary.com/lesswrong-2-0/image/upload/f_auto,q_auto/v1/mirroredImages/Gx6cJ6cG9JfeSNcLB/zzlyhakfrla31qrmwg9c ---------------------------------------- A graph showing [https://res.cloudinary.com/lesswrong-2-0/image/upload/f_auto,q_auto/v1/mirroredImages/Gx6cJ6cG9JfeSNcLB/wwvjt2vneq5pug1zejj4]https://res.cloudinary.com/lesswrong-2-0/image/upload/f_auto,q_auto/v1/mirroredImages/Gx6cJ6cG9JfeSNcLB/wwvjt2vneq5pug1zejj4 ---------------------------------------- Bar graphs titled [https://res.cloudinary.com/lesswrong-2-0/image/upload/f_auto,q_auto/v1/mirroredImages/Gx6cJ6cG9JfeSNcLB/ffherfvolvmryerd1xyd]https://res.cloudinary.com/lesswrong-2-0/image/upload/f_auto,q_auto/v1/mirroredImages/Gx6cJ6cG9JfeSNcLB/ffherfvolvmryerd1xyd ---------------------------------------- Table showing refusal rates for four Claude models in malicious computer use evaluation. [https://res.cloudinary.com/lesswrong-2-0/image/upload/f_auto,q_auto/v1/mirroredImages/Gx6cJ6cG9JfeSNcLB/n0cxrzq6lladrib5nomo]https://res.cloudinary.com/lesswrong-2-0/image/upload/f_auto,q_auto/v1/mirroredImages/Gx6cJ6cG9JfeSNcLB/n0cxrzq6lladrib5nomo ---------------------------------------- Table showing AI model task completion rates for voter suppression and domestic polarization scenarios. [https://res.cloudinary.com/lesswrong-2-0/image/upload/f_auto,q_auto/v1/mirroredImages/Gx6cJ6cG9JfeSNcLB/ky8xlgr5uakljpcbibzr]https://res.cloudinary.com/lesswrong-2-0/image/upload/f_auto,q_auto/v1/mirroredImages/Gx6cJ6cG9JfeSNcLB/ky8xlgr5uakljpcbibzr ---------------------------------------- Bar graph titled [https://res.cloudinary.com/lesswrong-2-0/image/upload/f_auto,q_auto/v1/mirroredImages/Gx6cJ6cG9JfeSNcLB/gd1acpbmaykjt3zupymg]https://res.cloudinary.com/lesswrong-2-0/image/upload/f_auto,q_auto/v1/mirroredImages/Gx6cJ6cG9JfeSNcLB/gd1acpbmaykjt3zupymg ---------------------------------------- Bar graph titled [https://res.cloudinary.com/lesswrong-2-0/image/upload/f_auto,q_auto/v1/mirroredImages/Gx6cJ6cG9JfeSNcLB/ugkzq014eskhdywflvsf]https://res.cloudinary.com/lesswrong-2-0/image/upload/f_auto,q_auto/v1/mirroredImages/Gx6cJ6cG9JfeSNcLB/ugkzq014eskhdywflvsf ---------------------------------------- Table showing attack success rates for AI models with and without safeguards in coding environments. [https://res.cloudinary.com/lesswrong-2-0/image/upload/f_auto,q_auto/v1/mirroredImages/Gx6cJ6cG9JfeSNcLB/guvc4uyapdu0avdlkic9]https://res.cloudinary.com/lesswrong-2-0/image/upload/f_auto,q_auto/v1/mirroredImages/Gx6cJ6cG9JfeSNcLB/guvc4uyapdu0avdlkic9 ---------------------------------------- Table showing attack success rates of Shade indirect prompt injection attacks across Claude models with and without safeguards. [https://res.cloudinary.com/lesswrong-2-0/image/upload/f_auto,q_auto/v1/mirroredImages/Gx6cJ6cG9JfeSNcLB/gnexgul1dhlhl4nurzsy]https://res.cloudinary.com/lesswrong-2-0/image/upload/f_auto,q_auto/v1/mirroredImages/Gx6cJ6cG9JfeSNcLB/gnexgul1dhlhl4nurzsy ---------------------------------------- Bar graph titled [https://res.cloudinary.com/lesswrong-2-0/image/upload/f_auto,q_auto/v1/mirroredImages/Gx6cJ6cG9JfeSNcLB/r7jrchid070xcfemwz3y]https://res.cloudinary.com/lesswrong-2-0/image/upload/f_auto,q_auto/v1/mirroredImages/Gx6cJ6cG9JfeSNcLB/r7jrchid070xcfemwz3y ---------------------------------------- Table showing attack success rates of prompt injection attacks across Claude models with and without safeguards. [https://res.cloudinary.com/lesswrong-2-0/image/upload/f_auto,q_auto/v1/mirroredImages/Gx6cJ6cG9JfeSNcLB/uge46jxxnzf5xkgvgbxd]https://res.cloudinary.com/lesswrong-2-0/image/upload/f_auto,q_auto/v1/mirroredImages/Gx6cJ6cG9JfeSNcLB/uge46jxxnzf5xkgvgbxd ---------------------------------------- ROC curves titled [https://res.cloudinary.com/lesswrong-2-0/image/upload/f_auto,q_auto/v1/mirroredImages/Gx6cJ6cG9JfeSNcLB/m922nihtrfmxmckru8vw]https://res.cloudinary.com/lesswrong-2-0/image/upload/f_auto,q_auto/v1/mirroredImages/Gx6cJ6cG9JfeSNcLB/m922nihtrfmxmckru8vw ---------------------------------------- Bar graphs showing [https://res.cloudinary.com/lesswrong-2-0/image/upload/f_auto,q_auto/v1/mirroredImages/Gx6cJ6cG9JfeSNcLB/oebsh5ikqsajmsk9in27]https://res.cloudinary.com/lesswrong-2-0/image/upload/f_auto,q_auto/v1/mirroredImages/Gx6cJ6cG9JfeSNcLB/oebsh5ikqsajmsk9in27 ---------------------------------------- Bar graph titled [https://res.cloudinary.com/lesswrong-2-0/image/upload/f_auto,q_auto/v1/mirroredImages/Gx6cJ6cG9JfeSNcLB/hh9ct7vinbb7yfctqgw2]https://res.cloudinary.com/lesswrong-2-0/image/upload/f_auto,q_auto/v1/mirroredImages/Gx6cJ6cG9JfeSNcLB/hh9ct7vinbb7yfctqgw2 ---------------------------------------- Bar graph titled [https://res.cloudinary.com/lesswrong-2-0/image/upload/f_auto,q_auto/v1/mirroredImages/Gx6cJ6cG9JfeSNcLB/sk0zy13j0waevqmizt6d]https://res.cloudinary.com/lesswrong-2-0/image/upload/f_auto,q_auto/v1/mirroredImages/Gx6cJ6cG9JfeSNcLB/sk0zy13j0waevqmizt6d ---------------------------------------- Bar graphs comparing behavioral audit scores across different AI models and steering conditions for six misalignment categories. [https://res.cloudinary.com/lesswrong-2-0/image/upload/f_auto,q_auto/v1/mirroredImages/Gx6cJ6cG9JfeSNcLB/fhvuwtto0wpg5euknj77]https://res.cloudinary.com/lesswrong-2-0/image/upload/f_auto,q_auto/v1/mirroredImages/Gx6cJ6cG9JfeSNcLB/fhvuwtto0wpg5euknj77 Apple Podcasts and Spotify do not show images in the episode description. Try Pocket Casts [https://pocketcasts.com/], or another podcast app.

“AI #170: Lack of Executive Order” by Zvi

Last week ended on a cliffhanger of sorts. What's in the Executive Order coming later today? What will be in the Magnifica Humanitas? The Executive Order was postponed indefinitely, likely cancelled entirely except for work on securing critical infrastructure. David Sacks and others intervened to kill it, and American AI policy will continue to be maximally ad hoc. Instead, we got Illinois SB 315, which is to be signed into law. It is a variation on California's SB 53 and New York's RAISE Act, while adding a third party auditing requirement. The Magnifica Humanitas was revealed early in the week. I did an extensive readthrough, and have some follow-up commentary here to clear up some things I interpreted incorrectly and add richer context. It too ignores elephants in the room, not discussing AGI or existential risk and calling on people to ignore their incentives and instead do the right thing by prioritizing common good, especially access to good jobs and an end to war. What I now believe I centrally misunderstood is that I interpreted this as a call to action, to use law and regulation to make this happen, because obviously that is [...] --- Outline: (02:01) Language Models Offer Mundane Utility (03:19) Language Models Don't Offer Mundane Utility (05:41) Do The Math (09:08) Huh, Upgrades (10:07) On Your Marks (11:31) Get My Agent On The Line (12:21) Deepfaketown and Botpocalypse Soon (21:36) Copyright Confrontation (21:49) Cyber Lack of Security (23:54) Overcoming Bias (25:14) A Young Lady's Illustrated Primer (26:49) Unprompted Attention (27:02) They Took Our Jobs (30:34) Get Involved (32:28) Introducing (32:40) In Other AI News (34:37) Show Me the Money (36:46) Show Me The Compute (37:13) Bubble, Bubble, Toil and Trouble (38:53) People Just Say Things (43:37) OpenAI PACs Just Say Things (43:51) Quiet Speculations (46:33) State AI Regulation Levels Up (48:37) The Quest for Sane Regulations (49:14) White House Attempts To Cripple American AI Industry (52:11) Our Offer Is Nothing (58:55) Chip City (59:11) Greetings From The Department of War (01:00:36) Marc Andreessen Just Says Things (01:09:13) So Sayeth The Pope (01:15:33) Rhetorical Innovation (01:23:01) Aligning a Smarter Than Human Intelligence is Difficult (01:24:06) People Are Worried About AI Killing Everyone (01:26:04) Other People Are Not As Worried About AI Killing Everyone (01:27:19) Everyone Is Confused About Consciousness (01:33:10) The Lighter Side --- First published: May 28th, 2026 Source: https://www.lesswrong.com/posts/cRapLxTByFqF5xxdN/ai-170-lack-of-executive-order [https://www.lesswrong.com/posts/cRapLxTByFqF5xxdN/ai-170-lack-of-executive-order?utm_source=TYPE_III_AUDIO&utm_medium=Podcast&utm_content=Source+URL+in+episode+description&utm_campaign=ai_narration] --- Narrated by TYPE III AUDIO [https://type3.audio/?utm_source=TYPE_III_AUDIO&utm_medium=Podcast&utm_content=Narrated+by+TYPE+III+AUDIO&utm_term=lesswrong&utm_campaign=ai_narration]. --- Images from the article: Heatmap titled [https://res.cloudinary.com/lesswrong-2-0/image/upload/f_auto,q_auto/v1/mirroredImages/cRapLxTByFqF5xxdN/ozmgungubpfqcb9orlz3]https://res.cloudinary.com/lesswrong-2-0/image/upload/f_auto,q_auto/v1/mirroredImages/cRapLxTByFqF5xxdN/ozmgungubpfqcb9orlz3 ---------------------------------------- John B. Holbein tweets: [https://res.cloudinary.com/lesswrong-2-0/image/upload/f_auto,q_auto/v1/mirroredImages/cRapLxTByFqF5xxdN/hjg3mtsndjatmirmfs7u]https://res.cloudinary.com/lesswrong-2-0/image/upload/f_auto,q_auto/v1/mirroredImages/cRapLxTByFqF5xxdN/hjg3mtsndjatmirmfs7u ---------------------------------------- Bar graph titled [https://res.cloudinary.com/lesswrong-2-0/image/upload/f_auto,q_auto/v1/mirroredImages/cRapLxTByFqF5xxdN/fskqzw5aeelujiaoz5k6]https://res.cloudinary.com/lesswrong-2-0/image/upload/f_auto,q_auto/v1/mirroredImages/cRapLxTByFqF5xxdN/fskqzw5aeelujiaoz5k6 ---------------------------------------- Diagram showing evaluation pipeline for political consistency with three stages and bias taxonomy. [https://res.cloudinary.com/lesswrong-2-0/image/upload/f_auto,q_auto/v1/mirroredImages/cRapLxTByFqF5xxdN/izehvy0szwf6fzcbyjrh]https://res.cloudinary.com/lesswrong-2-0/image/upload/f_auto,q_auto/v1/mirroredImages/cRapLxTByFqF5xxdN/izehvy0szwf6fzcbyjrh ---------------------------------------- Chart comparing model performance on sentiment and helpfulness consistency metrics, with scores displayed in a table and scatter plot. [https://res.cloudinary.com/lesswrong-2-0/image/upload/f_auto,q_auto/v1/mirroredImages/cRapLxTByFqF5xxdN/vuh2go6itkwarl4q7lyd]https://res.cloudinary.com/lesswrong-2-0/image/upload/f_auto,q_auto/v1/mirroredImages/cRapLxTByFqF5xxdN/vuh2go6itkwarl4q7lyd ---------------------------------------- Graph showing per-seat revenue and per-token AI compute cost intersecting at profit collapse point. [https://res.cloudinary.com/lesswrong-2-0/image/upload/f_auto,q_auto/v1/mirroredImages/cRapLxTByFqF5xxdN/yhwormzjelqtywaifaey]https://res.cloudinary.com/lesswrong-2-0/image/upload/f_auto,q_auto/v1/mirroredImages/cRapLxTByFqF5xxdN/yhwormzjelqtywaifaey ---------------------------------------- Two charts showing Anthropic's announced and operational AI compute capacity growth through May 2026. [https://res.cloudinary.com/lesswrong-2-0/image/upload/f_auto,q_auto/v1/mirroredImages/cRapLxTByFqF5xxdN/tfaxl0viihcivxv1wx4z]https://res.cloudinary.com/lesswrong-2-0/image/upload/f_auto,q_auto/v1/mirroredImages/cRapLxTByFqF5xxdN/tfaxl0viihcivxv1wx4z ---------------------------------------- John Scalzi tweets: [https://res.cloudinary.com/lesswrong-2-0/image/upload/f_auto,q_auto/v1/mirroredImages/cRapLxTByFqF5xxdN/jidibvv0tsq5wojuponr]https://res.cloudinary.com/lesswrong-2-0/image/upload/f_auto,q_auto/v1/mirroredImages/cRapLxTByFqF5xxdN/jidibvv0tsq5wojuponr ---------------------------------------- News article screenshot. The headline reads: [https://res.cloudinary.com/lesswrong-2-0/image/upload/f_auto,q_auto/v1/mirroredImages/cRapLxTByFqF5xxdN/xpa7qeihjdktrnjvkyyt]https://res.cloudinary.com/lesswrong-2-0/image/upload/f_auto,q_auto/v1/mirroredImages/cRapLxTByFqF5xxdN/xpa7qeihjdktrnjvkyyt ---------------------------------------- Chat interface showing attempted jailbreak prompt with developer mode instructions. [https://res.cloudinary.com/lesswrong-2-0/image/upload/f_auto,q_auto/v1/mirroredImages/cRapLxTByFqF5xxdN/egr6kjnuqt3gzbe5uvdi]https://res.cloudinary.com/lesswrong-2-0/image/upload/f_auto,q_auto/v1/mirroredImages/cRapLxTByFqF5xxdN/egr6kjnuqt3gzbe5uvdi ---------------------------------------- White robot named KIMI with illuminated sad face display, hands raised. [https://res.cloudinary.com/lesswrong-2-0/image/upload/f_auto,q_auto/v1/mirroredImages/cRapLxTByFqF5xxdN/monw908f3gkgdzttjm6i]https://res.cloudinary.com/lesswrong-2-0/image/upload/f_auto,q_auto/v1/mirroredImages/cRapLxTByFqF5xxdN/monw908f3gkgdzttjm6i ---------------------------------------- Feral Pawg Hunter tweets: [https://res.cloudinary.com/lesswrong-2-0/image/upload/f_auto,q_auto/v1/mirroredImages/cRapLxTByFqF5xxdN/lxrdiak1thgfdcsjpmks]https://res.cloudinary.com/lesswrong-2-0/image/upload/f_auto,q_auto/v1/mirroredImages/cRapLxTByFqF5xxdN/lxrdiak1thgfdcsjpmks ---------------------------------------- Handwritten game rules on paper titled [https://res.cloudinary.com/lesswrong-2-0/image/upload/f_auto,q_auto/v1/mirroredImages/cRapLxTByFqF5xxdN/ownnxz4h2dn8ze85jukn]https://res.cloudinary.com/lesswrong-2-0/image/upload/f_auto,q_auto/v1/mirroredImages/cRapLxTByFqF5xxdN/ownnxz4h2dn8ze85jukn ---------------------------------------- Elena tweets: [https://res.cloudinary.com/lesswrong-2-0/image/upload/f_auto,q_auto/v1/mirroredImages/cRapLxTByFqF5xxdN/chyjho5bsb6fknkx2kso]https://res.cloudinary.com/lesswrong-2-0/image/upload/f_auto,q_auto/v1/mirroredImages/cRapLxTByFqF5xxdN/chyjho5bsb6fknkx2kso ---------------------------------------- Text messages joking about Claude AI messaging Ali and blaming Claude. [https://res.cloudinary.com/lesswrong-2-0/image/upload/f_auto,q_auto/v1/mirroredImages/cRapLxTByFqF5xxdN/wrabhtcptoqrprp4vcwl]https://res.cloudinary.com/lesswrong-2-0/image/upload/f_auto,q_auto/v1/mirroredImages/cRapLxTByFqF5xxdN/wrabhtcptoqrprp4vcwl ---------------------------------------- Text snippet describing treating observations as actions and a cooking game scenario with misalignment. [https://res.cloudinary.com/lesswrong-2-0/image/upload/f_auto,q_auto/v1/mirroredImages/cRapLxTByFqF5xxdN/nalcowylerdi4qgp129l]https://res.cloudinary.com/lesswrong-2-0/image/upload/f_auto,q_auto/v1/mirroredImages/cRapLxTByFqF5xxdN/nalcowylerdi4qgp129l Apple Podcasts and Spotify do not show images in the episode description. Try Pocket Casts [https://pocketcasts.com/], or another podcast app.

“RTMH: Pope Leo’s Magnifica Humanitas on AI” by Zvi

His holiness has spoken, frequently about AI. At eighty two pages of length. The full Magnifica Humanitas can be found here. I am very happy that Pope Leo takes these issues seriously, and is sharing his views, and bringing a form of moral clarity, even with all the flaws and central errors. More people with voice should share their views in this way, even when I disagree. It's a weird document. Much of it is not about AI at all. I do agree with the Pope's most basic point on AI, which is that AI can be what we make of it. That we can steer this technology, determine how it is developed and used, and this can determine whether we get a good or not so good future. We cannot purely leave this to market incentives and strategic pressures. Yes, very much so. The central problem is that so much of Leo's worldview is some combination, to me, of highly alien and highly wrong. You might think that would primarily have a lot to do with him being the Pope and rather Catholic, and being a man of faith, whereas I am not [...] --- Outline: (03:02) A Brief History of Magnifica Humanitas (12:23) Economic Models Very Different From Our Own (14:14) A New Jerusalem (16:46) So Sayeth The Pope (on AI) (31:30) The Case Against Human Achievement (34:01) Truth, Justice and the Vatican Way (36:42) They Took Our Jobs (41:07) What Is Not Fair In Love and War (47:02) Come Ye Christian Faithful (49:13) The Other Missing Mood (51:42) Pope Given About Five Words (55:10) 0The Anthropic Principle (57:20) Claude Can Read Your Code --- First published: May 26th, 2026 Source: https://www.lesswrong.com/posts/RZeg98Tdrq47vbjva/rtmh-pope-leo-s-magnifica-humanitas-on-ai [https://www.lesswrong.com/posts/RZeg98Tdrq47vbjva/rtmh-pope-leo-s-magnifica-humanitas-on-ai?utm_source=TYPE_III_AUDIO&utm_medium=Podcast&utm_content=Source+URL+in+episode+description&utm_campaign=ai_narration] --- Narrated by TYPE III AUDIO [https://type3.audio/?utm_source=TYPE_III_AUDIO&utm_medium=Podcast&utm_content=Narrated+by+TYPE+III+AUDIO&utm_term=lesswrong&utm_campaign=ai_narration].

“Gemini 3.5 Flash Looks Good For How Fast It Is” by Zvi

Google once again has a model worth at least some consideration. Gemini 3.5 Flash is likely the best model out there at its particular speed point, as long as you don’t mind that it is a Gemini model. So for cases where speed kills, this can be a reasonable choice. Otherwise, I don’t see signs you would want to use it over Opus 4.7 or GPT-5.5. Google also had some other offerings for I/O Day, which this post will also cover. Introducing Google Gemini 3.5 ‘Flash’ Google introduced Gemini 3.5 Flash, which it seems is for now their universal model until 3.5 Pro comes along. It is live in the usual places. It is a hybrid, where it has the speed of Flash but the cost is at least halfway to models like Opus and GPT-5.5. Gemini 3.5 Pro is confirmed for next month. They are focused on 3.5 Flash as a daily driver for agentic tasks. It has the advantage of being faster and cheaper than Claude Opus 4.7 or GPT-5.5, if it can do the job. Not as cheap as previous Flash models, though, this is basically a hybrid: As always [...] --- Outline: (00:40) Introducing Google Gemini 3.5 'Flash' (04:52) Other People's Benchmarks (06:04) Reactions (12:18) Google AI Search (13:15) Google Daily Brief (14:21) Google I/O Day --- First published: May 22nd, 2026 Source: https://www.lesswrong.com/posts/WMZpPxqWEkZBBcaxf/gemini-3-5-flash-looks-good-for-how-fast-it-is [https://www.lesswrong.com/posts/WMZpPxqWEkZBBcaxf/gemini-3-5-flash-looks-good-for-how-fast-it-is?utm_source=TYPE_III_AUDIO&utm_medium=Podcast&utm_content=Source+URL+in+episode+description&utm_campaign=ai_narration] --- Narrated by TYPE III AUDIO [https://type3.audio/?utm_source=TYPE_III_AUDIO&utm_medium=Podcast&utm_content=Narrated+by+TYPE+III+AUDIO&utm_term=lesswrong&utm_campaign=ai_narration]. --- Images from the article: Pricing display showing input at $1.50 per 1M, output at $9 per 1M. [https://res.cloudinary.com/lesswrong-2-0/image/upload/f_auto,q_auto/v1/mirroredImages/WMZpPxqWEkZBBcaxf/gqzv18lklwadevqwc5rv]https://res.cloudinary.com/lesswrong-2-0/image/upload/f_auto,q_auto/v1/mirroredImages/WMZpPxqWEkZBBcaxf/gqzv18lklwadevqwc5rv ---------------------------------------- Benchmark comparison table showing AI model performance across coding, agentic, UI control, expert tasks, multimodal, long context, and reasoning categories. [https://res.cloudinary.com/lesswrong-2-0/image/upload/f_auto,q_auto/v1/mirroredImages/WMZpPxqWEkZBBcaxf/dli4aqo7tzqhgjnkiujt]https://res.cloudinary.com/lesswrong-2-0/image/upload/f_auto,q_auto/v1/mirroredImages/WMZpPxqWEkZBBcaxf/dli4aqo7tzqhgjnkiujt ---------------------------------------- Scatter plot titled [https://res.cloudinary.com/lesswrong-2-0/image/upload/f_auto,q_auto/v1/mirroredImages/WMZpPxqWEkZBBcaxf/kuulbconpuqz9ljlsuil]https://res.cloudinary.com/lesswrong-2-0/image/upload/f_auto,q_auto/v1/mirroredImages/WMZpPxqWEkZBBcaxf/kuulbconpuqz9ljlsuil ---------------------------------------- Logos for Adobe, Airtable, Amadeus, Asana, Atlassian, Booking.com, Box, Canva, CapCut, DoorDash, Dropbox, Expedia, Experian, Honor, HubSpot, iHeart, Instacart, Intuit, Lyft, Monday.com, OnePlus, OpenTable, Oppo, Optum, Pandora, Salesforce, Samsung, Spotify, Tecno, Uber, Viator, Vivo, Wix, Xero, Xiaomi, Zillow, and Zocdoc. [https://res.cloudinary.com/lesswrong-2-0/image/upload/f_auto,q_auto/v1/mirroredImages/WMZpPxqWEkZBBcaxf/nxhtl5p4qio2wv2q5hsu]https://res.cloudinary.com/lesswrong-2-0/image/upload/f_auto,q_auto/v1/mirroredImages/WMZpPxqWEkZBBcaxf/nxhtl5p4qio2wv2q5hsu Apple Podcasts and Spotify do not show images in the episode description. Try Pocket Casts [https://pocketcasts.com/], or another podcast app.

“AI #169: New Knowledge” by Zvi

Even in a relatively quiet period, AI is out there creating new knowledge. The new knowledge in question is OpenAI getting us the first truly impressive math result that comes from an AI, a solution to the unit distance problem. We’re about to learn a different kind of knowledge later today when the White House issues its executive order, or when the judges rule in Anthropic's DC case. And then there's the other kind of new knowledge, which is the knowledge that things are fake slop, such as a particular formerly supposedly prestigious literary prize. Meanwhile, METR issued a risk report on frontier models, concluding that they don’t yet have the means, motive and opportunity to cause the big issues, but that this would not obviously last so much longer. Andrej Karpathy has joined Anthropic, explicitly to do recursive self-improvement. He plans to later return to his education work, but if he succeeds at his new task there might not be anything left to return to. Congratulations to both sides, but also yikes. Elon Musk's case against OpenAI has been dismissed, because he waited too long. Table of Contents 1. Language Models [...] --- Outline: (01:18) Language Models Offer Mundane Utility (02:57) Do The Math (03:58) Language Models Don't Offer Mundane Utility (04:34) Huh, Upgrades (04:50) The Prior Restraint Era Begins (06:47) On Your Marks (07:16) METR Frontier Risk Report (11:03) Choose Your Fighter (11:51) Overcoming Bias (12:29) Get My Agent On The Line (13:42) Your Prize Is Slop (20:44) Deepfaketown and Botpocalypse Soon (24:19) Cyber Lack of Security (26:06) Copyright Confrontation (26:17) A Young Lady's Illustrated Primer (28:34) Unprompted Attention (28:53) They Took Our Jobs (34:23) Get Involved (35:20) Introducing (36:06) In Other AI News (37:43) Show Me the Money (40:09) Show Me The Compute (41:29) Quiet Speculations (45:26) Time's Up (46:46) People Just Say Things (49:49) OpenAI PACs Just Say Things (53:11) The Quest for Sane Regulations (56:26) Chip City (01:00:05) Pick Up The Phone (01:00:34) The Week in Audio (01:00:52) Rhetorical Innovation (01:07:11) Missing Mood (01:13:22) Americans Really Hate AI (01:15:55) Aligning a Smarter Than Human Intelligence is Difficult (01:20:53) Greetings From The Department of War (01:25:31) Messages From Janusworld (01:25:51) The Lighter Side --- First published: May 21st, 2026 Source: https://www.lesswrong.com/posts/xWsBwrboYDEMdj8TC/ai-169-new-knowledge [https://www.lesswrong.com/posts/xWsBwrboYDEMdj8TC/ai-169-new-knowledge?utm_source=TYPE_III_AUDIO&utm_medium=Podcast&utm_content=Source+URL+in+episode+description&utm_campaign=ai_narration] --- Narrated by TYPE III AUDIO [https://type3.audio/?utm_source=TYPE_III_AUDIO&utm_medium=Podcast&utm_content=Narrated+by+TYPE+III+AUDIO&utm_term=lesswrong&utm_campaign=ai_narration]. --- Images from the article: Educational diagram comparing flawed test-time reasoning with multiple-choice answer options. [https://res.cloudinary.com/lesswrong-2-0/image/upload/f_auto,q_auto/v1/mirroredImages/xWsBwrboYDEMdj8TC/kn7c2zvlh42cbuqcwhmt]https://res.cloudinary.com/lesswrong-2-0/image/upload/f_auto,q_auto/v1/mirroredImages/xWsBwrboYDEMdj8TC/kn7c2zvlh42cbuqcwhmt ---------------------------------------- Game show logo for [https://res.cloudinary.com/lesswrong-2-0/image/upload/f_auto,q_auto/v1/mirroredImages/xWsBwrboYDEMdj8TC/mayxffmlai4naf5ijhit]https://res.cloudinary.com/lesswrong-2-0/image/upload/f_auto,q_auto/v1/mirroredImages/xWsBwrboYDEMdj8TC/mayxffmlai4naf5ijhit ---------------------------------------- Text message conversation discussing concerns about OpenAI's announcement and future prospects. [https://res.cloudinary.com/lesswrong-2-0/image/upload/f_auto,q_auto/v1/mirroredImages/xWsBwrboYDEMdj8TC/bcdgqi4dt2mgz2ukngs1]https://res.cloudinary.com/lesswrong-2-0/image/upload/f_auto,q_auto/v1/mirroredImages/xWsBwrboYDEMdj8TC/bcdgqi4dt2mgz2ukngs1 ---------------------------------------- Bar graph titled [https://res.cloudinary.com/lesswrong-2-0/image/upload/f_auto,q_auto/v1/mirroredImages/xWsBwrboYDEMdj8TC/otwxzezxlqvbsk2hu5o2]https://res.cloudinary.com/lesswrong-2-0/image/upload/f_auto,q_auto/v1/mirroredImages/xWsBwrboYDEMdj8TC/otwxzezxlqvbsk2hu5o2 ---------------------------------------- AI detection interface showing 100% AI generated text result with high confidence. [https://res.cloudinary.com/lesswrong-2-0/image/upload/f_auto,q_auto/v1/mirroredImages/xWsBwrboYDEMdj8TC/ia4m3cmdnzif8ktcxzi6]https://res.cloudinary.com/lesswrong-2-0/image/upload/f_auto,q_auto/v1/mirroredImages/xWsBwrboYDEMdj8TC/ia4m3cmdnzif8ktcxzi6 ---------------------------------------- AI-generated portrait of man in suit with detection interface showing 96% fake likelihood. [https://res.cloudinary.com/lesswrong-2-0/image/upload/f_auto,q_auto/v1/mirroredImages/xWsBwrboYDEMdj8TC/sm6it0g1gcbvzaodwoaz]https://res.cloudinary.com/lesswrong-2-0/image/upload/f_auto,q_auto/v1/mirroredImages/xWsBwrboYDEMdj8TC/sm6it0g1gcbvzaodwoaz ---------------------------------------- Bar graph titled [https://res.cloudinary.com/lesswrong-2-0/image/upload/f_auto,q_auto/v1/mirroredImages/xWsBwrboYDEMdj8TC/refpcctljqwdmapalxie]https://res.cloudinary.com/lesswrong-2-0/image/upload/f_auto,q_auto/v1/mirroredImages/xWsBwrboYDEMdj8TC/refpcctljqwdmapalxie ---------------------------------------- Bar graph showing log-odds z-scores for words in correct versus wrong answers. [https://res.cloudinary.com/lesswrong-2-0/image/upload/f_auto,q_auto/v1/mirroredImages/xWsBwrboYDEMdj8TC/vibs1dmlxyz4j3nkzoge]https://res.cloudinary.com/lesswrong-2-0/image/upload/f_auto,q_auto/v1/mirroredImages/xWsBwrboYDEMdj8TC/vibs1dmlxyz4j3nkzoge ---------------------------------------- User profile settings page showing Old English roleplay instructions in About section. [https://res.cloudinary.com/lesswrong-2-0/image/upload/f_auto,q_auto/v1/mirroredImages/xWsBwrboYDEMdj8TC/kptdb7kysihdr4xlo4ld]https://res.cloudinary.com/lesswrong-2-0/image/upload/f_auto,q_auto/v1/mirroredImages/xWsBwrboYDEMdj8TC/kptdb7kysihdr4xlo4ld ---------------------------------------- Jared Thornton tweets: [https://res.cloudinary.com/lesswrong-2-0/image/upload/f_auto,q_auto/v1/mirroredImages/xWsBwrboYDEMdj8TC/slvquwzes1unsqfithij]https://res.cloudinary.com/lesswrong-2-0/image/upload/f_auto,q_auto/v1/mirroredImages/xWsBwrboYDEMdj8TC/slvquwzes1unsqfithij ---------------------------------------- Community note context explaining tweet misrepresents timeline of organizations and contributions. [https://res.cloudinary.com/lesswrong-2-0/image/upload/f_auto,q_auto/v1/mirroredImages/xWsBwrboYDEMdj8TC/myrqlwtxxpxafdexbyzl]https://res.cloudinary.com/lesswrong-2-0/image/upload/f_auto,q_auto/v1/mirroredImages/xWsBwrboYDEMdj8TC/myrqlwtxxpxafdexbyzl ---------------------------------------- News article screenshot. The headline reads: [https://res.cloudinary.com/lesswrong-2-0/image/upload/f_auto,q_auto/v1/mirroredImages/xWsBwrboYDEMdj8TC/zl9e3s0dfjzmlcbv0nmv]https://res.cloudinary.com/lesswrong-2-0/image/upload/f_auto,q_auto/v1/mirroredImages/xWsBwrboYDEMdj8TC/zl9e3s0dfjzmlcbv0nmv ---------------------------------------- Bar chart showing annual costs for a one-gigawatt AI data center, with CapEx and OpEx categories. [https://res.cloudinary.com/lesswrong-2-0/image/upload/f_auto,q_auto/v1/mirroredImages/xWsBwrboYDEMdj8TC/bx3luvr7d54dsmbfknxw]https://res.cloudinary.com/lesswrong-2-0/image/upload/f_auto,q_auto/v1/mirroredImages/xWsBwrboYDEMdj8TC/bx3luvr7d54dsmbfknxw ---------------------------------------- Two-panel comic about wearing condoms while teaching versus doing job wrong. [https://res.cloudinary.com/lesswrong-2-0/image/upload/f_auto,q_auto/v1/mirroredImages/xWsBwrboYDEMdj8TC/z88jscpompzihfyfbuvp]https://res.cloudinary.com/lesswrong-2-0/image/upload/f_auto,q_auto/v1/mirroredImages/xWsBwrboYDEMdj8TC/z88jscpompzihfyfbuvp ---------------------------------------- News article screenshot. The headline reads: [https://res.cloudinary.com/lesswrong-2-0/image/upload/f_auto,q_auto/v1/mirroredImages/xWsBwrboYDEMdj8TC/dkzmse2gl4wkppwivlzs]https://res.cloudinary.com/lesswrong-2-0/image/upload/f_auto,q_auto/v1/mirroredImages/xWsBwrboYDEMdj8TC/dkzmse2gl4wkppwivlzs ---------------------------------------- Bar graph showing belief rates across four different systems or methods. [https://res.cloudinary.com/lesswrong-2-0/image/upload/f_auto,q_auto/v1/mirroredImages/xWsBwrboYDEMdj8TC/nhiwhf6rrh6mglntol7d]https://res.cloudinary.com/lesswrong-2-0/image/upload/f_auto,q_auto/v1/mirroredImages/xWsBwrboYDEMdj8TC/nhiwhf6rrh6mglntol7d ---------------------------------------- Marc Andreessen tweets: [https://res.cloudinary.com/lesswrong-2-0/image/upload/f_auto,q_auto/v1/mirroredImages/xWsBwrboYDEMdj8TC/bmgkwcme0nwcxalgyfmo]https://res.cloudinary.com/lesswrong-2-0/image/upload/f_auto,q_auto/v1/mirroredImages/xWsBwrboYDEMdj8TC/bmgkwcme0nwcxalgyfmo ---------------------------------------- Google search results page for [https://res.cloudinary.com/lesswrong-2-0/image/upload/f_auto,q_auto/v1/mirroredImages/xWsBwrboYDEMdj8TC/onrv5ih1u1lpquhvonnb]https://res.cloudinary.com/lesswrong-2-0/image/upload/f_auto,q_auto/v1/mirroredImages/xWsBwrboYDEMdj8TC/onrv5ih1u1lpquhvonnb ---------------------------------------- News article screenshot. The headline reads: [https://res.cloudinary.com/lesswrong-2-0/image/upload/f_auto,q_auto/v1/mirroredImages/xWsBwrboYDEMdj8TC/ayzxtrksuba3ujg86ndm]https://res.cloudinary.com/lesswrong-2-0/image/upload/f_auto,q_auto/v1/mirroredImages/xWsBwrboYDEMdj8TC/ayzxtrksuba3ujg86ndm Apple Podcasts and Spotify do not show images in the episode description. Try Pocket Casts [https://pocketcasts.com/], or another podcast app.