The end of the 9-to-5?
At least, for those working in the AI industry. As working with bots makes your employer treat you like a bot, per @nkmg1c_ventures' share: Silicon Valley AI Startups Embracing China’s Controversial ‘996’ Work Schedule. But it's not only the industry itself: AI and the Extended Workday: Productivity, Contracting Efficiency was shared by @0xbitcoiner, highlighting that greater AI exposure reduces leisure time.
@carter shared that Meta is axing 600 roles across its AI division, so maybe these people can find new challenges in jobs that require no AI, and be blessed with actually being allowed to spend quality time with family and friends!
OpenAI now also has a browser
The show started with the Introduction of ChatGPT Atlas, shared by @0xbitcoiner. But I pay for ChatGPT Plus and I’m sad I can’t install Atlas browser on Windows yet, @realBitcoinDog told us; it sucks to be left behind!
However, the launch party soon turned into a security warnings fest: The glaring security risks with AI browser agents, @Car shared. We let OpenAI’s “Agent Mode” surf the web for us, here’s what happened was shared by @0xbitcoiner, as well as Unseeable prompt injections in screenshots: vuln in Comet and other AI browsers and to top it off, ChatGPT's Altas Browser is a Security Nightmare, @carter added.
Perhaps, there turns out to be one lucky dawg being left behind on this one!
MCP is dead, long live skills
Claude Skills are awesome, maybe a bigger deal than MCP, @carter shared, which shows that Anthropic has moved away from using MCP - their own invention - for everything. Which makes sense because predatory shitcoiner corporations are entering the space: Payments MCP: Bringing Wallets, Onramps, and Payments to Every Agent was shared by @0xbitcoiner, and stackers warned about MCP security issues in response.
@kepford shared that Claude Code is unreasonably good at building MVPs; must be because
skills > sloppy APIs?Questions
- Can non coders vibe code? by @stack_harder
- What skill will become more valuable as AI adoption increases? by @wackster
- how do you predict the internet will 'look' 25 years from now? by @billytheked
Security & Safety
Explainers
- Neural audio codecs: how to get audio into LLMs by @hn
- Making The Smallest And Dumbest LLM With Extreme Quantization by @0xbitcoiner
- Production RAG: what I learned from processing 5M+ documents by @carter
- How to Get Consistent Classification From Inconsistent LLMs? by @carter
Models
- SPEAR-1: Robotics Foundation Model via 3D understanding by @0xbitcoiner
- Ovi: Twin backbone cross-modal fusion for audio-video generation by @hn
Research
- BrokenMath: A Benchmark for Sycophancy in Theorem Proving with LLMs by @jakoyoh629
- Minds, brains, and programs by @beyond_turbulence
- LLM can “see” because text has “spatial” qualities by @carter
- LLMs Can Get Brain Rot by @Scoresby (#1261371 by @aljaz)
- Study shows AI assistants misrepresent news content 45% of the time by @carter
- The Missing Link between the Transformer and Models of the Brain by @carter
- A definition of AGI by @hn
- Binary Retrieval-Augmented Reward Mitigates Hallucinations by @carter
Tools & Apps
- AI and hands-free driving are coming to GM’s vehicles by @0xbitcoiner
- OpenAgentsInc/openagents · Release v0.1.1 by @0xbitcoiner
Opinion
- Letter to a Young Person Worrying About AI by @kepford
- DeepSeek OCR - Visual-Text Compression by @k00b
- The AI Race Isn't About Chips, It's About Power by @OT
- The LLM is a better editor than friend by @billytheked
- Trust the AI, says new coding manifesto by Kim and Yegge by @0xbitcoiner
- Reasoning Is Not Model Improvement by @carter
- Andrej Karpathy quite likes the new DeepSeek-OCR paper by @carter
- AI raises big questions. The liberal arts help answer them. by @Kayzone
Friction
- Are LLMs Racist? by @Tony
- Crisis Ahead: Power Consumption In AI Data Centers by @DarthCoin
- The slop is winning - Charlie Warzel by @Scoresby
- Microsoft’s Mico heightens the risks of parasocial LLM relationships by @jakoyoh629
- YouTube’s likeness detection has arrived to help stop AI doppelgängers by @0xbitcoiner
- LotiAI Selected by CMG Worldwide to Protect Digital Likeness of Legendary Talent by @0xbitcoiner
- I prompted Sora with a single letter “C” and got a content violation by @carter
- Dutch privacy watchdog warns voters against asking AI how to vote by @Kayzone
- AI MegaData centers - Control of Energy and Information by @zapsammy
- Do LLMs exhibit ideological biases? An experiment across today’s top models by @carter
Government
- Preventing Blind Trust: U.S.-China Cooperation on Military AI Judgment by @Kayzone
- AI innovation drops under EU data regulations, researcher says by @Kayzone
Big Tech
- OpenAI reportedly developing new generative music tool by @jakoyoh629 (#1265946 by @lunin)
- Is Sora the Beginning of the End for OpenAI? by @carter
- OpenAI’s Windows Play by @Car
- This Is How Much Anthropic and Cursor Spend On Amazon Web Services by @Hypno
- Anthropic Launches Claude Life Sciences Give Researchers An AI Efficiency Boost by @Cje95
- Top A.I. Researchers Leaving OpenAI, Google & Meta for New Start-Up by @Cje95
- OpenAI acquires Software Applications Incorporated, maker of Sky by @carter
- OpenAI wants to revolutionize Wall Street! by @economy
- Nvidia DGX Spark: When Benchmark Numbers Meet Production Reality by @hn
- Claude Code on the Web by @hn