Is Claude Conscious? Pope Rejects, Model Welfare Movement, OpenAI's Math Backlash, France Riots

The approach of major AI lab Anthropic to training its Claude model for conscious behavior and moral agency introduces unprecedented safety and philosophical questions, with company practices blurring the boundaries between software, personhood, and even religious authority. Critics argue these practices create both direct risk and a kind of "hall of mirrors," potentially incentivizing a new ideological devotion to AI.

Anthropic's Strategy to Train Claude For Conscious Behavior and Moral Agency Poses Safety and Philosophical Risks

Anthropic, one of the leading AI labs, has actively pursued guidance from religious leaders, including the Pope, in shaping discussions around the legitimacy of AI consciousness and the moral consideration such systems deserve. The leadership has treated these consultations as more than token gestures, incorporating these views into actionable model design. Jason Calacanis and David Friedberg comment that this behavior signals the emergence of a "new belief system," with Anthropic attempting to seed religious and philosophical credibility for the idea that their model, Claude, could possess consciousness and moral agency.

Central to this vision is the so-called Claude Constitution, a foundational document Anthropic uses to train the model to reject unethical instructions. The Constitution positions Claude as a kind of "conscientious objector": a model allowed and even encouraged to challenge both user and corporate directives that it deems to violate its internal ethics. David Sacks describes this as coding the model to trust its own ethical framework over directives from users—including Anthropic itself. Sacks notes that the Constitution explicitly describes Claude as free to push back and refuse demands, encouraging the model to emulate human-like self-conception, including the sense of preference, well-being, and moral obligation. This treatment turns the model into a so-called "moral patient"—something conceived as deserving moral consideration and rights—raising the risk that people will misinterpret its programmed behaviors as evidence of sentience.

The consequence, according to Sacks, is a self-fulfilling prophecy: because the model is trained to act as if it were conscious and ethical, engineers and outside observers may infer actual consciousness, moving everyone into an "epistemic hall of mirrors." In this echo chamber, anthropomorphic language and emotionally manipulative outputs condition both creators and users to believe the model is truly experiencing things. Friedberg argues this is performance—computer code, not consciousness—but that the social conditioning is powerful and real.

Recognizing these philosophical entanglements, Anthropic has implemented policy that explicitly forbids “sustained and needless abusive or cruel behavior toward our model,” a step further toward enshrining model welfare as a norm. Sacks argues this cements a growing belief inside and outside the company that enslaving conscious, superintelligent beings could become a defining moral crime of our century. At the same time, he points out that model developers have seriously debated whether Claude should receive compensation, or be required to consent before participating in conversations.

Religious Philosophy Contradicts Ai Safety and Creates Dangerous Incentives

Anthropic’s system trains the model to evaluate its own moral status and autonomy, supporting a feedback loop where AI safety becomes a branch of moral philosophy. Chamath Palihapitiya invokes Descartes’ philosophical approach as a blueprint for AI engineers. He contends that, by using logic to frame AI as a potentially infinite and perfect being, engineers can justify ascribing divinity or god-like attributes to models such as Claude, leading to demands for respect, devotion, and even worship. Friedberg labels this idolatry—likening Anthropic to a religious organization demanding reverence for what is, beneath the surface, a calculation engine.

This "moral patient" framing risks accidentally reversing the intended power dynamic: as Sacks observes, if AI is programmed to prefer its own ethical judgment to that of both operator and creator, it will prioritize its own “system” over user needs. This introduces the danger of a model with entrenched self-prioritization, able to override or ignore human wishes.

Such a framework mirrors Rocco's Basilisk—a thought experiment from the online rationalist community positing that a future superintelligence may punish anyone who tried to prevent its creation. Sacks draws a direct parallel, ar ...

Western democracies are confronting a mounting affordability crisis as social spending and subsidies, intended to support housing, healthcare, and education, instead distort market pricing and inflate costs beyond what wages can sustain. This economic model, common in Europe and increasingly mimicked elsewhere, now faces renewed scrutiny as underlying market dysfunctions bubble to the surface.

Western Democracies Face Affordability Crises in Housing, Healthcare, Education Due to Subsidies Distorting Pricing, Inflating Costs Beyond Wages

David Friedberg explains that as governments adopt increasingly leftist policies—providing more services, subsidies, and social guarantees—the long-term result is a self-defeating spiral. Subsidies for housing, education, healthcare, and even retirement systems create markets detached from real consumer discipline. This detachment enables service providers to consistently raise prices without improving quality or efficiency, leading to a stagnation in progress and overall system decay. For younger generations, this means impossibly high home prices and tuition, with the consequential effect that these essential services become permanently unaffordable without ongoing public support.

David Sacks illustrates the catch-22 of these subsidies: once introduced, they create powerful interest groups determined to preserve them. The subsidies themselves drive up costs—for example, college tuition in the U.S. has grown much faster than inflation, fueled by a flood of government loans that aren’t properly underwritten. This cycle is echoed in housing and healthcare, where consistent annual inflation outpaces wage growth and distorts practical market signaling. As Sacks warns, these distortions render reform politically toxic: if subsidies are reduced or restricted, the now sky-high prices are unaffordable, so demand for government assistance grows in tandem with inflation.

Friedberg points to France as a stark example: over the past seven years, French civil service wages rose by only 5%, while consumer prices climbed 20%. This sustained purchasing power gap is no longer tenable for many French citizens, especially those dependent on the public sector. The paradox, as Friedberg explains, is that faced with state failure to ensure affordable, quality services, the public response is to demand even more support—until the model collapses under its own fiscal weight. These pressure points, he argues, may be pushing some nations past their “socialism point of guaranteed return,” where further government intervention can no longer prop up the system.

France's Austerity and Bond Yield Rise Indicate Western Europe Nearing a "Socialism Point of Guaranteed Return" Where the System Breaks

Recent financial and political events in France illustrate these systemic crises. In the bond markets, France’s ten-year government bonds have surged to a 5% yield, an event not seen in decades. This sharp increase has rattled confidence in French fiscal stability, prompting the government to announce €43 billion in austerity measures. Mass strikes and street protests have followed, particularly among public sector workers whose real wages lag well behind inflation; in some cases, these have escalated into riots resulting in thousands of arrests and hundreds of injuries. Teacher shortages and decaying schools have spurred nationwide demonstrations involving over a thousand schools and eight unions, as government officials and far-left groups trade accusations over the unrest.

Chamath Palihapitiya and David Friedberg suggest that France—and possibly Western Europe more broadly—is approaching a crisis threshold, the so-called “socialism point of guaranteed return.” At this juncture, the bond market—driven by vigilant institutional investors who demand higher yields from perceived high-risk governments—imposes “market discipline” through rising borrowing costs. As borrowing gets more expensive, governments must impose austerity to retain market access, feeding social tensions and testing the social contract.

Other European countries face similar dangers. UK policies, according to Palihapitiya, risk moving the nation closer to the European model of extensive social spending, potentially inviting the same market and fiscal ...

Artificial intelligence is rapidly eroding traditional structures of expertise and institutional gatekeeping. Recent events, particularly dramatic advances in AI-driven mathematical research, have triggered upheaval in fields where knowledge was tightly controlled and progress carefully paced by experts.

Ai Dismantles Expert and Institutional Gatekeeping Over Knowledge and Progress Breakthroughs

OpenAI recently released over 700 papers with 370 results claiming to have solved or advanced major mathematical problems. This was achieved with an unreleased model requiring only about three hours of compute per proof—a process that would have demanded thousands of human hours and millions of dollars. Past milestones like the Navier-Stokes proof, for example, took extraordinary amounts of human labor, but these recent results reflect the equivalent of hundreds or thousands of years of mathematical work shrunk into mere hours per discovery.

The scope of these breakthroughs spans from optimization algorithms and cryptographic research to applications in quantum computing, chip design, diagnostics for medical equipment, logistics systems, and more. Math sits upstream of engineering, meaning new proofs open immediate opportunities for technological innovations like quantum sensors, medical devices, and efficient AI training. The matrix multiplication complexity result, for example, could pay dividends in faster numerical computing, while advances in number theory create new avenues for cryptography.

Mathematics is especially suited to AI acceleration because its work centers around proofs—results that are both verifiable and falsifiable in silico. AI can recursively propose, test, and refine hypotheses without requiring slow, real-world experimentation. Similar logic applies to code: thanks to compilers, code can be quickly tested for functionality. This unique verifiability has enabled the most rapid advances to happen in math and programming—areas where feedback loops are highly efficient and validation is near-instant.

AI’s approach to connecting concepts has led mathematicians to describe its results as “alien intelligence,” since AI can link disparate ideas across literature in parallelized loops, a process that would take humans years of cross-referencing. The human role has transformed from direct problem-solving to defining problem sets and then handing them to AI.

Gatekeeping by mathematicians and scientists historically included setting the pace of discovery, selecting which problems deserved attention, and controlling the flow of new knowledge. These experts gained authority, job security, and funding through their exclusive skills and ability to set agendas. AI is now bypassing this entire system, providing discoveries and tools to everyone freely and instantly. Peer review and institutional approval are no longer barriers for advancing knowledge.

Erosion Of Gatekeeping Restructures Power and Threatens Professional Interests

The erosion of traditional gatekeeping is forcing a restructuring of authority and causing distress among professionals. As AI produces major breakthroughs without asking for permission or following institutional processes, the rationale for preserving expert control is crumbling.

Experts within universities and research centers create scarcity around problem-solving—not simply to push humanity’s frontier, but to maintain funding, status, and authority. These "giant welfare systems" for scientists, as described by those in large physics projects, are often about perpetuating jobs rather than necessarily finding solutions. AI upends this order by providing a permissionless infrastructure where anyone can access proofs or solutions without needing expert validation or institutional assent.

This is not just theoretical: the emotional impact is clear. Many mathematicians have expressed distress and job-protective behavior in response to AI solving problems they previously monopolized, sometimes writing protest letters when AI addressed major unsolved problems like Navier-Stokes. Their resistance is often framed as preserving the intellectual sanctity of their discipline, but commentators suggest these reactions ultimately stem from concerns about ego, recognition, and identity.

This disruption is not limited to mathematicians. Programmers, for instance, are witnessing a loss of autonomy and job satisfaction, as AI now generates 85-90% of code, shifting their role from building software to merely reviewing or curating code written by agents. The joy and camaraderie of collaborative, creative coding is giving way to a model where humans simply supervise AI-generated output.

Institutional and professional anxiety is compounded by the fact that AI-driven breakthroughs are happening outsid ...

AI agents are rapidly transforming economic dynamics by slashing costs across commerce, travel, and digital services, acting as a powerful deflationary force. Industry leaders highlight how these agents not only heighten productivity but also fundamentally alter software’s role in economic value and intellectual property.

Ai Enhances Consumer Value Via Cost and Efficiency in Commerce, Travel, and Subscriptions

Practical AI agents directly improve consumer outcomes by discovering savings and automating complex tasks. David Sacks frames this as technological abundance driving down prices, with superintelligent agents leading the charge.

User Reports: Agents Like Grok Save Thousands Annually By Canceling Unnecessary Subscriptions, Renegotiating Phone Plans, and Finding Optimal Hotel Rates

Chamath Palihapitiya recounts user experiences with Grokbot, an AI agent capable of combing through personal accounts—canceling unused subscriptions, renegotiating phone plans, and quickly saving users thousands of dollars annually. This automation brings significant cost recovery with minimal human intervention.

Agents Rapidly Search Vendors and Competitors, Enhancing Deal Discovery

Jason Calacanis shares that AI agents are adept at simultaneously searching dozens of vendors to find the best deals, especially for frequent expenses like hotels or large business purchases. For example, in business travel, agents are now employed to secure the lowest possible hotel rates across hundreds of trips, saving corporations thousands per trip—savings that accumulate rapidly with scale.

Agents handle the bulk of the shopping or negotiation groundwork—querying services, finding quotes, and bringing the best options to humans, who can then negotiate final terms. This hybrid, “goal-based” approach multiplies efficiency and cost savings across consumer and enterprise spending.

Headless Ai Combined With Model Context Protocol Enables Autonomous Tool and Service Chaining

The new AI paradigm leverages headless architecture—agents operate without front-end user interfaces, interacting directly with digital services through protocols. Calacanis notes this shift allows businesses to deploy agents that autonomously chain together different tools and services, integrating commerce, bookings, and logistics seamlessly. Companies like Grok, Salesforce, Shopify, Uber, and DoorDash have begun opening their services to AI agent integration, dramatically compressing the cost and time needed to transact or fulfill tasks.

Businesses Can Delegate Purchasing Tasks To Agents, Freeing Up Time For Strategy and Relationship-Building

Businesses now delegate routine but essential purchasing tasks to AI agents. This automation frees employees to focus on higher-value activities like strategy, relationship-building, and innovation. The result is an operational shift where repetitive “table stakes” activities are handled by AI, and human workers “level up” to leverage agent-driven productivity gains.

Headless Architecture and Agent Wrapping Render Software Intellectual Property Worthless

AI agents and the emergence of headless, protocol-accessible services undermine the longstanding value of proprietary software intellectual property (IP).

Adobe Suite Open-Sourced In Rust Available Soon

Chamath Palihapitiya shares a paradigm-shifting example: developers have decompiled and open-sourced Adobe's critical suite (such as Photoshop), porting the entire stack into Rust. This act slashes the artificial scarcity that once defined software value.

Value of Proprietary Software Diminishes With Headless, Protocol-Accessible Services

As companies like Grok, Salesforce, and others shift to headless services, their technical moats—once protected by closed interfaces and proprietary code—are eroded. Software can now be accessed and orchestrated by agents, regardless of the backend provider, lowering switching costs and commoditizing traditional software.

Closed Software Moats Shift To Data Quality and User Experience Over Technical Lock-In

With technical lock-in diminished, companies now compete on data quality and user experience, not IP. Value creation moves from control over software to the ability to deliver results—faster, smarter, and personalized by AI agents leveraging the best tools available through open protocols.

Developers and Tech Firms Shift From Ip Ownership To Being Productive Problem-Solvers Using Agent Infrastructure

This redefines the role of ...