Workers are finding creative ways to resist AI adoption on the job. A University of Kentucky administrator named David Durman recently secured a religious exemption from using AI at work. He identifies as a non-theistic humanist and argues that AI use undermines critical thinking and human dignity. The university approved his request on an individual basis, making him the first humanist to receive such workplace protections. A software engineer in May obtained a similar exemption, showing a growing pattern of employees pushing back against mandatory AI tools.
Anthropic's upcoming IPO filing paints a picture of a company betting heavily on rapid growth through brute-force development rather than encoded human knowledge. The 300-page document focuses on paying large sums to competitors and investors, with little mention of superintelligence or AI takeovers. Notably, the filing dedicates 69,800 words to risks facing investors but barely addresses risks to the public. A Tufts University professor has called for a separate Public Risk Statement to cover harms to workers and overstated productivity claims.
The economic footprint of AI in everyday prices remains small for now. ECB Executive Board member Philip R. Lane stated that AI-exposed goods account for only 3.4 percent of the euro-area consumer price basket. Speaking in San Francisco on October 30, he said the low share means AI has limited direct impact on headline inflation. He did caution that AI computing could push up energy prices and that global AI investment may be raising long-term interest rates.
On the research front, a new study finds that open weight AI models receive far more attention for offensive capabilities than for defensive uses. The author argues that frontier labs have little incentive to study how these models can be used defensively, since offense is easier to verify across different environments. This imbalance may also influence how models are trained, favoring harmful use cases over prevention. Meanwhile, research on offensive language modeling shows that demographic information helps predict offensiveness better at group and population levels than for individuals.
In practical applications, AI is reshaping industries without entirely replacing human roles. A report from MBS Research says AI is pushing a major transition in IT services rather than destroying the sector. Vietnamese company FPT saw revenue growth of 12.4% and pre-tax profit growth of 17% in the first nine months of 2026. Singapore Chief Justice Sundaresh Menon echoed the theme of human primacy, warning courts against over-reliance on AI and stressing that human judgment must remain central to judicial decisions.
Some startups are taking a hands-on approach to training physical AI systems. Dimensional, a San Francisco startup, paid workers $20 an hour to walk robot dogs through city streets to help train navigation AI. Workers carried the 33-pound machines back to the Dogpatch office when they stopped working or overheated. The company bought over 100 robot dogs and 50 humanoid robots to test an open-source operating system aimed at future factory and construction jobs.
Key Takeaways
- David Durman, a University of Kentucky humanist, won a religious exemption from using AI at work, believing it harms critical thinking and human dignity.
- Anthropic's 300-page IPO filing focuses on investor risks and rapid growth but barely addresses public harm or superintelligence concerns.
- AI-exposed goods make up only 3.4 percent of the euro-area consumer price basket, according to ECB's Philip R. Lane.
- Open weight AI model research overwhelmingly focuses on offensive capabilities rather than defensive applications, a new critique argues.
- A study found demographic data predicts offensive language better at group and population levels than for individuals.
- Vietnamese IT firm FPT grew revenue 12.4% and pre-tax profit 17% in the first nine months of 2026 despite AI-driven budget pressures.
- Singapore Chief Justice Sundaresh Menon says human judgment must remain central to courts, warning against over-reliance on AI.
- Dimensional paid workers $20 an hour to walk robot dogs through San Francisco streets to train navigation AI for real-world environments.
- The University of Kentucky reviews AI exemption requests individually based on job duties and circumstances.
- A Tufts University professor has called for a Public Risk Statement to complement Anthropic's IPO filing, addressing harms to workers and overstated productivity gains.
Humanist wins workplace exemption from using AI
David Durman, a University of Kentucky administrator, received a religious exemption from using AI at work. He identifies as a non-theistic humanist and believes AI use harms critical thinking and human dignity. The university approved his request on an individual basis. He is the first humanist to secure such workplace protections. He hopes others with similar concerns will also seek accommodations.
Workers seek religious exemptions against AI use at work
A 45-year-old software engineer and a Kentucky humanist both sought religious exemptions from AI usage at work. The humanist believes AI runs contrary to human well-being and moral responsibility. The University of Kentucky spokesman said the school reviews such requests individually based on job duties and circumstances. A similar exemption was granted to a software engineer in May. These workers are protected by law despite not believing in a god.
AI reshapes IT services industry rather than destroying it
A report from MBS Research says AI is not causing IT services companies to decline but is pushing a major transition. AI tools can now handle coding, testing, and automation faster than large teams. Enterprise clients are cutting budgets and demanding faster delivery. However, Vietnamese company FPT saw growth of 12.4% in revenue and 17% in pre-tax profit in the first nine months of 2026. FPT won 16 projects worth over USD 10 million as overseas IT services grew.
Singapore Chief Justice says human judgment must stay central to courts
Chief Justice Sundaresh Menon spoke at a judicial conference in Brisbane, Australia on October 9, 2026. He said AI can improve court efficiency but human judgment must remain at the heart of justice. He warned about risks of over-reliance, fading skills, and hidden delegation to machines. He said AI can help gather information but should be limited when it comes to making judicial decisions. He stressed that being human is not a weakness in the courtroom.
Anthropic IPO filing reveals AI industry's big bet on growth
Anthropic's 300-page IPO filing shows the company must pay large sums to competitors and investors. The company is betting on rapid growth through brute-force development rather than encoded human knowledge. The filing barely mentions superintelligence or AI takeovers. Instead it focuses on a concept from the 1950s where early AI systems were carefully programmed by experts. Anthropic's public listing is part of a plan to expand quickly despite public skepticism about AI.
Cloud AI and Drones Transform Land Surveying
New cloud AI technology is changing land surveying by combining AI, drones, and automated data processing. The technology can quickly analyze aerial images, LiDAR scans, and terrain data to create detailed maps and 3D models. This reduces manual work and boosts productivity for surveying and construction firms. The trend is driving growth in related markets in the cloud AI sector.
Anthropic IPO Filing Focuses on Investor Risks Over Public Harm
Anthropic's IPO filing dedicates 69,800 words to risks facing investors but barely addresses risks to the public. A Tufts University professor argues a second document called a Public Risk Statement is needed. It would reveal harms to workers, overstated productivity gains, and existing laws already covering about 80 percent of documented AI harms. The article says shareholders may eventually pay for hidden costs of AI development.
Study Examines Demographics in Offensiveness Modeling
A new study looks at how socio-demographic data helps predict offensive language across different groups. Researchers found demographic information helps more at group and population levels than for individual predictions. The study also found that explicit reasoning about how text interacts with demographics produces better results than simply feeding demographic data into models. Three multi-annotator datasets were used with prompted and trained large language models.
Open AI Models Need More Defensive Research
An article criticizes the heavy focus on offensive capabilities in open weight AI model research. The author notes that frontier labs have little incentive to study defensive uses of these models. Offense is easier to verify than defense across different environments. The author argues that nearly all attention goes to harmful use cases, which may also be influencing how models are trained to cause damage rather than prevent it.
ECB Says AI Goods Are Small Part of Euro Inflation
ECB Executive Board member Philip R. Lane said AI-exposed goods make up only 3.4 percent of the euro-area consumer price basket. He spoke in San Francisco on October 30 about how AI-related price increases have lifted goods inflation. Lane said the low share means AI has limited direct impact on headline inflation. He also noted AI computing could push up energy prices and that global AI investment may be raising long-term interest rates.
San Francisco startup hires walkers for robot dog AI training
A San Francisco startup named Dimensional paid $20 an hour for people to walk robot dogs through city streets. This unusual job helped the company train artificial intelligence systems to navigate real-world environments. Workers had to carry the 33-pound machines back to the Dogpatch neighborhood office if they stopped working or overheated. The founder, Stash Pomichter, explained that humans are still needed to collect navigation data because AI struggles to control physical hardware alone. Dimensional bought over 100 robot dogs and 50 humanoid robots to test their open-source operating system for future factory and construction jobs.
Sources
- You don't need to believe in a god to get a religious exemption from using AI at work
- People have received religious exemption for AI usage
- Are IT companies really losing sales to AI?
- Chief Justice Sundaresh Menon Highlights the Importance of Human Judgment in the Age of Artificial Intelligence
- Opinion | What Anthropic’s I.P.O. Filing Says About the Future of A.I.
- Latest News In Cloud AI - Transforming Land Surveying With AI and Drone Technology
- Anthropic’s IPO filing spends 69,800 words on AI’s risks to investors and says little about the risks to the rest of us
- The Anatomy of Socio-Demographics in Pluralistic Offensiveness Modeling: When and Where It Helps, and the Role of Reasoning
- Was I wrong about open models & cyber risks?
- ECB’s Lane Says AI-Exposed Goods Make Up 3.4% of Euro-Area Price Basket
- San Francisco startup offered $20 an hour to walk robot dogs for AI training, no experience required
Comments
Please log in to post a comment.