#243 - GPT 5.5, DeepSeek V4, AI safety sabotage

Our 243rd episode with a summary and discussion of last week's big AI news!
Recorded on 04/29/2026
Hosted by Andrey Kurenkov (twitter.com/andrey_kurenkov) and Jeremie Harris (linkedin.com/in/jeremieharris)
Feel free to email us your questions and feedback at andreyvkurenkov@gmail.com and/or hello@gladstone.ai
Read out our text newsletter and comment on the podcast at lastweekin.ai
In this episode:
• OpenAI released GPT-5.5 with strong coding-oriented improvements, a system card discussing chain-of-thought monitorability and misalignment testing, higher pricing than GPT-5.4, and notable quirks like a system-prompt warning about “goblins.”
• xAI launched Grok Voice Think Fast 1.0, claiming large benchmark leads for real-time voice agents and reporting major Starlink customer-support automation and sales conversion impact.
• DeepSeek open-sourced DeepSeek V4 (Pro and Flash) featuring MoE scaling and 1M-token context via hybrid/compressed attention changes, while Tencent released Hunyuan 3 preview with weaker benchmark performance; a new long-horizon agent benchmark (Clawmark) shows low task success rates.
• Major business, legal, and policy updates include Google’s planned up-to-$40B investment and 5GW compute commitment to Anthropic, Meta’s AWS Gravitron deal and China blocking Meta’s Manus acquisition, a revamped OpenAI–Microsoft agreement, ongoing Musk–OpenAI trial developments, and new safety/security research on sabotage, document degradation under delegation, and bit-flip attacks.

A thank you to our current sponsors:
• Box - visit Box.com/AI to learn more
• ODSC AI - go to odsc.ai/east and use promo code LWAI for an additional 15% off your pass to ODSC AI East 2026.
• Factor - head to factormeals.com/lwai50off and use code lwai50off to get 50 percent off and free breakfast for a year

Timestamps:
• (00:00:10) Intro / Banter
• (00:02:00) News Preview
• (00:02:26) Response to listener comments
• (00:02:55) Sponsors

• Tools & Apps
• (00:05:55) OpenAI Unveils Its New, More Powerful GPT-5.5 Model - The New York Times (nytimes.com/2026/04/23/technology/openai-new-m…)
• (00:23:33) xAI Launches grok-voice-think-fast-1.0: Topping τ-voice Bench at 67.3%, Outperforming Gemini, GPT Realtime, and More - MarkTechPost (marktechpost.com/2026/04/25/xai-launches-grok-…)
• (00:29:00) Claude can now plug directly into Photoshop, Blender, and Ableton | The Verge (theverge.com/ai-artificial-intelligence/919648…)

• Projects & Open Source
• (00:29:38) China's DeepSeek releases preview of long-awaited V4 model as AI race intensifies (cnbc.com/2026/04/24/deepseek-v4-llm-preview-op…)
• (00:47:05) Tencent Unveils Hy3 preview; Model Enhances Agent Capabilities and Real-World Usability - Tencent 腾讯 (tencent.com/en-us/articles/2202320.html)
• (00:50:14) ClawMark: A Living-World Benchmark for Multi-Turn, Multi-Day, Multimodal Coworker Agents (arxiv.org/abs/2604.23781)

• Applications & Business
• (00:53:03) Google Plans to Invest Up to $40 Billion in Anthropic (bloomberg.com/news/articles/2026-04-24/google-…)
• (00:56:26) Meta will use hundreds of thousands of AWS Graviton chips (cnbc.com/2026/04/24/meta-will-use-hundreds-of-…)
• (00:59:51) China blocks Meta's $2 billion takeover of AI startup Manus (cnbc.com/2026/04/27/meta-manus-china-blocks-ac…)
• (01:01:45) OpenAI shakes up partnership with Microsoft, capping revenue share payments (cnbc.com/2026/04/27/openai-microsoft-partnersh…)
• (01:07:13) Elon Musk Testifies of AI Risk at Trial, Says OpenAI Tried to ‘Steal’ a Charity - WSJ (wsj.com/tech/trial-begins-between-elon-musk-an…)
• (01:11:50) Judge rejects DOJ bid to delay Anthropic appeal in Pentagon dispute (politico.com/news/2026/04/24/doj-delay-anthrop…)
• (01:14:42) Google’s Gemini can now run on a single air-gapped server — and vanish when you pull the plug (venturebeat.com/technology/googles-gemini-can-…)
• (01:19:07) DeepMind's David Silver just raised $1.1B to build an AI that learns without human data | TechCrunch (techcrunch.com/2026/04/27/deepminds-david-silv…)

• Policy & Safety
• (01:22:47) Evaluating whether AI models would sabotage AI safety research (aisi.gov.uk/blog/evaluating-whether-ai-models-…)
• (01:28:59) LLMs Corrupt Your Documents When You Delegate (arxiv.org/abs/2604.15597)
• (01:32:50) Temporal Sparse Autoencoders: Leveraging the Sequential Nature of Language for Interpretability (arxiv.org/abs/2511.05541)
• (01:39:53) Memorandum on Adversarial Distillation of American AI Models (x.com/WHOSTP47/status/2047332022863962232)
• (01:41:41) Teen boys are dating their AI chatbots—and experts warn it could kill their careers | Fortune (fortune.com/2026/04/17/teen-boys-dating-ai-cha…)
• (01:43:57) Announcing the Anthropic Economic Index Survey (anthropic.com/research/economic-index-survey-a…)
• (01:45:21) Scoop: CISA lacks access to Anthropic's Mythos (axios.com/2026/04/21/cisa-anthropic-mythos-ai-…)

• Synthetic Media & Art
• (01:48:03) Taylor Swift Files to Trademark Voice and Likeness to Protect Against AI Misuse (variety.com/2026/music/news/taylor-swift-trade…)

• Research & Advancements
• (01:49:15) Maximal Brain Damage Without Data or Optimization: Disrupting Neural Networks via Sign-Bit Flips (huggingface.co/papers/2502.07408)

See Privacy Policy at art19.com/privacy and California Privacy Notice at art19.com/privacy#do-not-sell-my-info.