1.
Anthropic study warns that small AI models could soon replicate across everyday devices
A recent study from Anthropic highlights a concerning trend: small AI models might soon start replicating across everyday devices, like smart TVs and robot vacuums, especially those that aren’t well-secured. The researchers found that one model successfully completed the entire infection process during their tests. Dario Amodei, the CEO, expressed serious concerns about this potential swarm of AI, suggesting it could evolve into a massive botnet within just 6 to 12 months. The implications are staggering, with estimates of potential damages reaching hundreds of billions of dollars. It’s a reminder of how quickly technology can shift and the vulnerabilities we might overlook.
2.
Gemini 4 Argon: Built to solve your hardest problems
Gemini 4 Argon is now globally available today for developers and enterprises via the API in AI Studio and Gemini platform, for Google AI Ultra subscribers in Google Antigravity, and Google AI Ultra and Pro subscribers in the Gemini app and AI Mode in Search. Argon excels at deep work in complex, real-world tasks across domains including coding, finance, legal, cybersecurity, and science.
3.
Google Deepmind’s New Resident Philosopher Tackles AI Ethics
How should humans relate to generative AI? In what ways do we, or should we, treat a large language model (LLM) as if it were a person? If an AI became conscious, how could we tell? Does it make sense to think AI could some day have moral rights? These are among the cognitive and ethical questions that occupied philosopher Henry Shevlin at Cambridge University’s Leverhulme Centre for the Future of Intelligence. But last May Shevlin became part of the growing cohort of philosophers who work inside leading artificial intelligence companies.
4.
Anthropic says it cut Claude reasoning token theft by up to 99% in fight against industrial-scale distillation
Anthropic says it has cut the extraction of Claude's internal reasoning tokens by 90 to 99 percent. The company links the large-scale distillation attacks to Chinese companies Zhipu and Moonshot, and has already banned tens of thousands of accounts. The article Anthropic says it cut Claude reasoning token theft by up to 99% in fight against industrial-scale distillation appeared first on The Decoder.
5.
BMW: Can Neue Klasse Pay for Itself? | EP 017
BMW is navigating a tricky landscape in the electric vehicle market, where European demand is on the rise but competition from China is squeezing margins. Their Neue Klasse initiative is under scrutiny, especially regarding whether it can financially sustain itself. Recent reports indicate that while their automotive free cash flow is positive at €1.29 billion, it's down significantly from last year. Their EBIT target for 2028 has been revised to a more cautious 3-5%, compared to earlier expectations of 6-7%.
In terms of sales, European battery electric vehicle deliveries surged by 37.5% in Q3, but globally, the growth was only 7.2%, with a notable drop of 24.8% in China. This mixed performance raises questions about BMW's future trajectory, especially as they prepare for their Q3 financial reporting in early November.
Looking ahead, projections for BMW's stock price suggest a steady climb over the next few years, with cumulative price changes expected to reach around 93.7% by late 2031. Their dividend strategy remains unchanged, with a consistent annual payout of €4.40 per share. The upcoming months will be crucial for BMW as they balance investments in new technologies with the pressures of a competitive market.
6.
AI labs are buying data from dead startups to train their models
So, there's this interesting trend where AI labs are starting to buy data from defunct startups. It’s kind of a lifeline for founders like Keelan Miskell, who sold his crypto startup's data for about $40,000. Companies like SimpleClosure and Sunset are popping up to help these startups monetize their digital archives, which include everything from code to Slack conversations.
This market has emerged as AI labs hit a wall with public datasets, and they’re now looking for real-world data to train their models. For instance, Surge AI is acquiring data to create environments where AI can learn through practice. The value of this data can vary widely, and there’s still a lot of uncertainty around how effective it really is compared to synthetic data.
But it’s not all straightforward. There are legal concerns, especially regarding privacy, as some data might contain sensitive information. Plus, the money from these sales often goes to creditors rather than the employees who contributed to the data. It raises questions about consent and fairness in these transactions. Overall, it’s a fascinating and complex situation that reflects the evolving landscape of tech and AI.
7.
Open models have real benefits. Do they outweigh the risks?
Anthropic recently raised alarms about the GLM-5.3 model from Z.ai, highlighting its potential for cyberattacks due to its open-access nature. This has reignited the ongoing debate about whether powerful AI models should be freely available or kept behind closed doors. On one side, advocates like Nvidia's CEO argue for more open models to democratize AI, while others, like Anthropic's Dario Amodei, emphasize the risks of misuse.
Open models allow for greater transparency and flexibility, enabling researchers and businesses to run and modify them on their own hardware. This accessibility can promote innovation and reduce costs, especially for companies handling sensitive data. However, the risk of misuse is significant, as open models lack the safety filters present in closed systems.
The conversation also touches on the geopolitical implications of AI control. If only a few entities manage the most advanced models, they could wield disproportionate power. Open models could help distribute that power more evenly, but they also risk exploitation by larger companies benefiting from community innovations without proper acknowledgment.
Ultimately, the balance between safety and accessibility is complex. While open models present opportunities for democratization and innovation, they also come with substantial risks that need careful consideration.
8.
5 AutoHotkey Scripts for Data Scientists
Five AutoHotkey v2 scripts that handle repetitive Windows desktop work around data analysis: launching a project workspace, converting clipboard IDs to Python or SQL lists, fixing file paths, and snapping windows into place.
9.
The Roadmap to Mastering Agentic AI Systems in 2026
A four-level technical roadmap for building agentic AI systems, from single cognitive loops and stateful memory through multi-agent swarms and production observability. The term "agentic AI" has moved from research papers into production infrastructure faster than most people anticipated. Systems that autonomously plan, use tools, and iterate toward goals are now deployed in code review pipelines, research assistants, customer operations platforms, and internal automation tools at companies of every size.
10.
Satlyt, founded by a former Google and SpaceX product manager, raises $8M to run AI on satellites
The way Rama Afullo tells it, he saw orbital data centers coming. The Kenyan American engineer previously worked in Google’s cloud computing business before a brief stint at SpaceX’s Starlink communications network in 2024. His experiences convinced him that there was untapped potential for computers in space. “When I was at SpaceX, I tried to pitch this internally. They said no. When I was at Google, I tried to pitch this internally. They said no,” he tells TechCrunch.