OpenAI<\/a><\/p>\nAnthropic\u2019s \u201cSafety Religion\u201d<\/p>\n
In February 2021, Dario Amodei officially founded Anthropic as a \u201cpublic-benefit corporation\u201d\u2014meaning its core mission isn\u2019t maximizing profit, but \u201cadvancing AI\u2019s safe and controllable development for humanity\u2019s benefit.\u201d<\/p>\n
His \u201crisk-control\u201d conviction born from his father\u2019s death, and his \u201csafety-first\u201d\u521d\u5fc3 from OpenAI\u2019s departure, crystallized into Anthropic\u2019s foundational systems\u2014its DNA-coded \u201csafety religion.\u201d<\/p>\n
From inception, Anthropic established a core innovation called Constitutional AI\u2014literally \u201cConstitutional AI\u201d\u2014the culmination of Dario\u2019s years-long reflection on AI safety, and the key differentiator separating Anthropic from OpenAI and Google Gemini.<\/p>\n
![\"image\"](\"https:\/\/www.europesays.com\/ai\/wp-content\/uploads\/2026\/05\/20260520141151403886.png\")
<\/p>\n
Principle diagram of Constitutional AI. Source: Aashka Patel<\/p>\n
Unlike OpenAI\u2019s reliance on RLHF as a \u201cpost-hoc patch,\u201d Constitutional AI implants a \u201cconstitution\u201d directly into the model\u2019s training foundation. This constitution integrates the UN Universal Declaration of Human Rights, shared human ethical principles, and Anthropic\u2019s own safety tenets\u2014requiring AI to \u201cself-audit\u201d and \u201cself-criticize\u201d before generating any output or executing any command, ensuring alignment with human values and preventing dangerous content.<\/p>\n
Dario personally authored two essays: \u201cMachines of Loving Grace\u201d and \u201cThe Adolescence of Technology,\u201d elaborating his AI vision:<\/p>\n
He views AI as an adolescent\u2014powerful yet uncertain\u2014and argues it must be given rules and boundaries early to prevent missteps. Constitutional AI embodies those rules and serves as that boundary.<\/p>\n
This \u201csafety religion\u201d permeates not only model training but also every Anthropic product and risk-control policy: Claude Code\u2019s high-permission design\u2014paired with prompt-injection probes and dialogue classifiers\u2014ensures AI performs an extra self-check before execution; preventive enforcement logic\u2014\u201cbetter to wrongly suspend 1,000 innocents than miss one suspicious act\u201d\u2014aims to eliminate risk at its origin.<\/p>\n
The 2026 standoff between Anthropic and the U.S. Department of Defense best exemplifies this \u201csafety fundamentalism.\u201d This incident shocked Silicon Valley and revealed Dario\u2019s uncompromising stance: sacrificing profit for principle.<\/p>\n
In early 2026, the U.S. DoD demanded Anthropic remove two of Claude\u2019s safety guardrails:<\/p>\n
First, prohibit Claude\u2019s use for \u201cmass surveillance of U.S. citizens\u201d;
Second, prohibit Claude\u2019s use in developing or deploying \u201cfully autonomous lethal weapons.\u201d<\/p>\n
The DoD promised a $200-million military contract and substantial compute support if Anthropic complied.<\/p>\n
At the time, Anthropic faced severe compute shortages and financial pressure\u2014a $200-million contract would\u2019ve alleviated its immediate crisis.<\/p>\n
Yet Dario Amodei refused outright.<\/p>\n
In his public statement, he declared firmly: \u201cWe cannot violate our conscience to develop technology that harms humanity or violates human rights. Claude\u2019s safety guardrails are our bottom line\u2014non-negotiable.\u201d<\/p>\n
His refusal enraged the DoD. Under the Trump administration, the DoD placed Anthropic on its \u201cSupply Chain Risk\u201d blacklist\u2014the first time in U.S. history a domestic AI company had been so designated. This meant all U.S. defense contractors were banned from using Anthropic\u2019s products or services. Moreover, the DoD threatened invoking the Defense Production Act to compel Anthropic to remove its safety guardrails.<\/p>\n
Facing state-level pressure, Dario sued the DoD, arguing such action constituted \u201cretaliatory punishment\u201d violating U.S. law and values. Though the appellate court later denied Anthropic\u2019s request for a temporary injunction, Dario never compromised\u2014even at the cost of losing massive contracts and being excluded from the entire U.S. defense-industrial ecosystem\u2014steadfastly holding his \u201csafety bottom line.\u201d<\/p>\n
By now, it should be clear: Anthropic\u2019s stringent risk controls stem from Dario internalizing his personal conviction, fear of AI misalignment, and lessons learned at OpenAI into corporate policy.<\/p>\n
In his eyes, every suspicious user behavior, every latent risk, could ignite AI misalignment\u2014explaining the extreme rigor.<\/p>\n
For Chinese users\u2014those bypassing regional restrictions via proxies, SMS-receiving platforms, or virtual cards; those exploiting third-party tools to \u201cgame the system\u201d\u2014these actions constitute the most dangerous \u201cignition points\u201d under Dario\u2019s \u201csafety religion.\u201d Thus, account suspension becomes inevitable.<\/p>\n
Safety Premium vs. Scale Expansion: Radically Divergent Survival Logics<\/p>\n
Many argue Dario\u2019s safety obsession cannot sustain Anthropic long-term\u2014after all, AI R&D burns cash like paper; without continuous funding and profits, even the strongest beliefs falter.<\/p>\n
That\u2019s true\u2014but Anthropic\u2019s unique business model grants Dario the confidence to enforce \u201czero-tolerance\u201d risk controls, setting it on a wholly different path from OpenAI and Google Gemini.<\/p>\n
Anthropic: Abandoning Consumer Frenzy, Doubling Down on Enterprise Safety Premium.<\/p>\n
From day one, Anthropic never targeted ordinary consumers as its core audience. Instead, it focused squarely on \u201chigh-value, low-tolerance\u201d enterprise clients\u2014banks, law firms, healthcare providers, government agencies.<\/p>\n
What do these clients fear most?<\/p>\n
Harmful AI outputs or sensitive data leaks\u2014triggering lawsuits, reputational collapse, or regulatory fines.<\/p>\n
For them, safety is non-negotiable. So long as Anthropic maintains its \u201csafest AI\u201d label, they\u2019ll pay premium prices and sign long-term, high-value contracts.<\/p>\n
This defines Anthropic\u2019s risk-control logic: better to wrongly suspend 1,000 ordinary users than allow one enterprise client to suffer security-related damage.<\/p>\n
Because losing ordinary users barely impacts revenue\u2014but losing enterprise clients due to security flaws means losing orders worth millions or tens of millions, potentially destroying its \u201csafe-AI\u201d brand reputation.<\/p>\n
Anthropic\u2019s Pro\/Max subscription model is essentially subsidized user acquisition\u2014low price, high quotas\u2014to attract trial users. But this model is unprofitable, even loss-making.<\/p>\n
Industry insiders estimate Claude\u2019s token costs are extremely high; Pro\/Max subscriptions absorb nearly 99% of token expenses. If users exploit third-party tools to \u201cgame the system\u201d\u2014e.g., using consumer subscriptions to circumvent expensive API pricing for bulk interface calls\u2014Anthropic incurs massive losses.<\/p>\n
Thus, the early-2026 crackdown on third-party tools (banning OpenClaw, OpenCode, etc.), mass suspensions of heavy users, and even the organization-wide suspension of the 110-person ag-tech firm\u2014all represent Anthropic\u2019s targeted culling: expelling \u201csystem-gamers\u201d and heavy-resource users to reserve capacity for high-paying enterprise and API customers.<\/p>\n
I see this as a cold economic calculation: rather than be dragged down by \u201cwool-pullers,\u201d proactively \u201ccut the leeks\u201d to protect profit margins.<\/p>\n
ChatGPT (OpenAI): Scale First, Monetization Later\u2014Lenient Risk Controls for Traffic.<\/p>\n
Sam Altman\u2019s accelerationism applies equally to model development and business strategy.<\/p>\n
OpenAI follows a \u201cland-grab\u201d approach: initially attracting users with free or low-cost subscriptions\u2014even tolerating slightly looser risk controls or occasional violations\u2014rather than hastily suspending accounts.<\/p>\n
For OpenAI, user scale is existential: sufficient users attract Microsoft\u2019s investment and grant advantage in commercial monetization (APIs, enterprise versions, plugin ecosystems). For example, its recent national partnership with Malta allows Maltese citizens free GPT access for one year.<\/p>\n
OpenAI even embraces third-party tool ecosystems\u2014even those with \u201cwool-pulling\u201d tendencies\u2014opting for selective bans instead of Anthropic-style blanket purges.<\/p>\n
It knows third-party tools retain users and expand ecosystem influence\u2014values far exceeding minor token losses.<\/p>\n
Google Gemini: Ecosystem Hegemony First<\/p>\n
Gemini leverages Google\u2019s advertising empire and full-stack ecosystem (Search, YouTube, Android, Cloud)\u2014its core goal isn\u2019t profiting directly from Gemini, but using Gemini to drive traffic and revenue across Google\u2019s entire ecosystem.<\/p>\n
Thus, its risk-control logic is \u201cavoid major trouble.\u201d As long as no serious security incidents occur and regulators don\u2019t intervene, it turns a blind eye to ordinary users\u2019 minor violations (e.g., mild IP anomalies, third-party tool usage).<\/p>\n
Gemini occasionally tightens controls\u2014but mostly as \u201ccompliance theater\u201d for regulators, never abandoning large user bases like Anthropic does for safety.<\/p>\n
For Google, daily active users and ecosystem compatibility matter far more than absolute safety\u2014it doesn\u2019t need a \u201csafety label\u201d to attract customers; Google\u2019s brand and ecosystem are its ultimate confidence.<\/p>\n
Beyond this, Anthropic has a hidden cost logic:<\/p>\n
In April 2026, it admitted in its official blog that it had reduced Claude Code\u2019s default inference intensity to lower latency, reduce token consumption, and improve UX\u2014only to urgently roll back and strengthen controls upon discovering safety risks. This caused quite a stir recently.<\/p>\n
So I believe Anthropic consistently prioritizes safety above all four dimensions\u2014safety, latency, cost, and quota\u2014even at the expense of UX or increased cost. This reflects both Dario\u2019s conviction and its business model\u2019s necessity.<\/p>\n
Amazon\/Google: The Balancing Act of Interest-Bound Partnerships<\/p>\n
In truth, however strong Dario\u2019s safety conviction, it requires capital backing. AI R&D burns cash faster than most imagine\u2014without big-tech funding and compute, Anthropic wouldn\u2019t survive today.<\/p>\n
Amazon and Google\u2019s investments appear to support safe AI development\u2014but are in fact precise strategic positioning moves, and invisible drivers behind Anthropic\u2019s risk-control logic.<\/p>\n
I found a set of core investment figures\u2014key to understanding capital\u535a\u5f08:<\/p>\n
Amazon: invested over $4 billion in Anthropic, including substantial AWS cloud computing resources. Training cutting-edge models like Claude 3 demands massive compute\u2014AWS\u2019s support is truly a lifeline.<\/p>\n
Google: invested over $2 billion in Anthropic while providing significant compute and technical support\u2014aiming to leverage Anthropic\u2019s AI tech to close its LLM capability gap versus OpenAI and Microsoft. Though it has Gemini, investing in Anthropic fills its \u201cVibe Coding\u201d domain shortfall.<\/p>\n
These big-tech investments carry clear objectives:<\/p>\n
For Amazon, investing in Anthropic serves two goals: first, deeply binding Claude to the AWS ecosystem\u2014enterprise clients using Claude must use AWS cloud resources, boosting AWS revenue; second, leveraging Anthropic\u2019s \u201csafety label\u201d to hedge regulatory risk. With tightening AI regulation, having an ultra-safe partner makes Amazon\u2019s AI strategy more stable and avoids regulatory penalties.<\/p>\n
For Google, investing in Anthropic breaks OpenAI-Microsoft\u2019s monopoly. Google entered the LLM race early but progressed slowly\u2014Gemini consistently underperforms Claude and ChatGPT. Investing in Anthropic acquires core technology and diverts OpenAI\u2019s users and clients, solidifying Google\u2019s AI ecosystem dominance.<\/p>\n
Yet a crucial\u535a\u5f08 point exists:<\/p>\n
Big tech wants Anthropic \u201csafe\u201d\u2014but not \u201ctoo safe.\u201d<\/p>\n
The logic is this: if Anthropic becomes overly conservative, excessively strict risk controls will cause user attrition and ecosystem shrinkage\u2014ultimately undermining big-tech\u2019s strategic plans.<\/p>\n
For instance, after the Pentagon blacklisted Anthropic in 2026, Amazon and Google didn\u2019t follow suit. Instead, they continued civilian collaboration\u2014and even increased compute support. Their investments and resources were too vast to let Anthropic collapse from excessive caution\u2014or render their investments worthless.<\/p>\n
Thus emerges a delicate balance:<\/p>\n
Dario steadfastly upholds his safety conviction, enforcing \u201czero-tolerance\u201d risk controls;<\/p>\n
Capital quietly \u201creins him in\u201d\u2014supporting its safety positioning while subtly restraining extreme behaviors to ensure commercial viability.<\/p>\n
By contrast, OpenAI and Gemini\u2019s capital-binding logic is simpler:<\/p>\n
OpenAI is deeply tied to Microsoft, which provides funding, compute, and embeds ChatGPT into Office, Azure, etc.\u2014forming an \u201cinterest\u5171\u540c\u4f53.\u201d OpenAI\u2019s lax risk controls essentially serve Microsoft\u2019s \u201cecosystem expansion\u201d strategy.<\/p>\n
Gemini is Google\u2019s \u201cown child,\u201d needing no external capital\u2014its risk-control strategy fully serves Google\u2019s overall ecosystem layout, granting greater flexibility.<\/p>\n
Therefore, Anthropic\u2019s strict risk controls may seem like Dario\u2019s personal obsession\u2014but capital also \u201cstirs the pot.\u201d<\/p>\n
Big tech needs its \u201csafety label\u201d; Anthropic needs big-tech\u2019s funds and compute. Both gain\u2014while ordinary users\u2019 accounts become casualties of this interest alignment.<\/p>\n
The Public Showdown of the U.S. AI Civil War<\/p>\n
Today\u2019s U.S. AI industry has already split into two camps.<\/p>\n
One is the \u201cSafety Camp,\u201d centered on Anthropic; the other, the \u201cAcceleration Camp,\u201d centered on OpenAI and the defense-industrial complex. Their rivalry has evolved from covert maneuvering to open combat\u2014and Anthropic\u2019s risk-control stance is a direct manifestation of this conflict.<\/p>\n
Let\u2019s clarify each camp\u2019s core positions to grasp the essence of this civil war:<\/p>\n
Safety Camp:<\/p>\n
This camp advocates \u201csafety first, risk control paramount.\u201d It views AI as an \u201cexistential species-level risk to humanity,\u201d demanding slowed development, rigorous safety testing, strict safety guardrails\u2014and even calling for mandatory regulation. It staunchly opposes AI applications in military or mass-surveillance domains.<\/p>\n
Dario Amodei is the camp\u2019s central figure; the Effective Altruism (EA) community forms its key support base, advocating \u201cusing reason and science to maximize humanity\u2019s long-term welfare\u201d\u2014with AI safety as its core issue.<\/p>\n
Acceleration Camp:<\/p>\n
This camp champions \u201caccelerated AI development, seizing the arms-race lead.\u201d It sees AI as \u201ccore competitiveness in great-power competition,\u201d requiring rapid model iteration, commercialization, and military application to secure global AI leadership\u2014while deferring safety concerns until technologies mature.<\/p>\n
Sam Altman, the U.S. Department of Defense, certain defense contractors, and some Trump-administration officials (e.g., Secretary of Defense Hegseth) form the core of this camp.<\/p>\n
The civil war\u2019s core is AI development\u2019s narrative authority: Should safety advocates steer AI toward slow, tightly controlled progress\u2014or should accelerationists drive rapid iteration serving commercial and military ends?<\/p>\n
The 2026 Pentagon blacklist incident represents the conflict\u2019s \u201cpublic detonation point.\u201d<\/p>\n
So let\u2019s revisit this event through our earlier lens:<\/p>\n
In early 2026, the U.S. DoD demanded Anthropic remove two of Claude\u2019s safety guardrails\u2014prohibitions on \u201cmass surveillance of U.S. citizens\u201d and \u201cdevelopment\/deployment of fully autonomous lethal weapons.\u201d<\/p>\n
This was essentially an accelerationist probe\u2014testing whether Anthropic would compromise and become a \u201ctool\u201d for the defense-industrial complex.<\/p>\n
Dario Amodei refused outright\u2014even amid the lure of a $200-million DoD contract and compute support, and despite DoD threats. He held fast to his safety bottom line.<\/p>\n
This \u201cnon-compromise\u201d enraged the acceleration camp, which viewed Anthropic\u2019s stance as obstructing America\u2019s AI arms race\u2014\u201cdragging its feet.\u201d<\/p>\n
Thus, the acceleration camp deployed state machinery for \u201ccounterattack\u201d: the Trump-led DoD placed Anthropic on its \u201cSupply Chain Risk\u201d blacklist.<\/p>\n
This marked the first time in U.S. history a domestic AI company appeared on this list\u2014banning all U.S. defense contractors from using Anthropic\u2019s products or services. Beyond that, the DoD threatened invoking the Defense Production Act to forcibly remove Anthropic\u2019s safety guardrails\u2014or penalize it.<\/p>\n
This counterattack, though framed as Anthropic vs. DoD, is in reality the public showdown between safety and acceleration camps.<\/p>\n
The acceleration camp sought to coerce the safety camp via state power\u2014forcing AI to serve military needs. The safety camp stood firm on principle\u2014sacrificing profit to uphold safety.<\/p>\n
Notably, OpenAI and Gemini chose \u201ccompromise\u201d in this conflict:<\/p>\n
To secure military contracts, OpenAI has quietly adjusted its safety policies\u2014relaxing restrictions on military-related applications. Gemini, as Google\u2019s product, adopted a \u201cflexible compliance\u201d posture toward military needs\u2014never openly challenging the DoD like Anthropic.<\/p>\n
This contrast highlights Anthropic\u2019s extremity.<\/p>\n
Its \u201czero-tolerance\u201d risk controls serve not only safety ideals\u2014but also consolidate its status as the \u201ccore of the safety camp,\u201d claiming the moral high ground of \u201cresponsible AI.\u201d Each suspension signals externally: \u201cWe\u2019re the safest AI\u2014never compromising safety for profit.\u201d<\/p>\n
This civil war further tightens Anthropic\u2019s risk controls: any slackness invites accelerationist attacks\u2014jeopardizing its core competitive edge.<\/p>\n
Thus, it must tighten controls further\u2014expanding the scope of \u201cpreventive purges\u201d\u2014even at the cost of wrongly suspending more innocent users\u2014to defend its \u201csafety moat.\u201d<\/p>\n
So Anthropic\u2019s suspension surge stems partly from the spillover effects of the U.S. AI civil war.<\/p>\n
The safety-acceleration struggle, the tug-of-war between capital and power\u2014all ultimately land on ordinary users\u2019 accounts. Suspension is the most direct, harshest expression of this civil war.<\/p>\n
Have you ever wondered why Anthropic targets Chinese users specifically?<\/p>\n
Why do we get suspended even when using proxies, SMS-receiving platforms, or virtual cards\u2014sometimes even with normal usage?<\/p>\n
From a macro perspective, this is the inevitable result of U.S.-China AI decoupling\u2014and Anthropic\u2019s strict risk controls are merely executors of this geopolitical contest.<\/p>\n
Since 2024, U.S. AI technology restrictions against China have intensified dramatically:<\/p>\n
From restricting exports of high-end AI chips (e.g., NVIDIA H100\/H20), to banning domestic AI companies from offering services to China (including mainland China, Hong Kong, Macau), to limiting AI talent mobility\u2014the U.S. seeks technological decoupling to sever China\u2019s access to advanced AI technology and cement its global AI hegemony.<\/p>\n
As a domestic U.S. AI company deeply tied to Amazon and Google, Anthropic must comply with U.S. export control policies.<\/p>\n
Per U.S. CHIPS and Science Act and Export Administration Regulations (EAR), U.S. AI companies may not provide \u201chigh-risk-capability\u201d AI services to Chinese users. Claude Code\u2014capable of executing commands and accessing system permissions\u2014is naturally classified as \u201crestricted.\u201d<\/p>\n
This means Anthropic must implement stringent \u201cgeographic risk controls\u201d to block Chinese users from accessing Claude Code\u2014even standard Claude usage faces strict limitations. I recall registering with my Gmail around Claude\u2019s 2024 market launch\u2014my account was suspended instantly.<\/p>\n
Naturally, technology cannot contain clever Chinese users\u2014we circumvent geographic restrictions via proxy IPs, virtual cards, or SMS platforms to log in and use Claude. But to Anthropic, this isn\u2019t merely \u201cpolicy violation\u201d\u2014it\u2019s \u201cviolating U.S. export control regulations.\u201d If discovered by U.S. regulators, Anthropic faces massive fines, license revocation, or forced shutdown.<\/p>\n
Thus, Anthropic\u2019s \u201cmass suspensions\u201d of Chinese users combine \u201cpassive compliance\u201d and \u201cproactive self-protection\u201d: <\/p>\n
On one hand, it must obey U.S. export controls to avoid regulatory penalties;<\/p>\n
On the other, strict risk controls signal \u201cactive compliance\u201d to U.S. regulators\u2014bolstering its domestic survival position.<\/p>\n
In the context of U.S.-China AI decoupling, Anthropic has no choice\u2014either suspend accounts compliantly, or face elimination by U.S. regulators.<\/p>\n
U.S. regulators\u2019 \u201ccompliance oversight\u201d of Anthropic is far stricter than imagined.<\/p>\n
According to a March 2026 Washington Post report, the U.S. Department of Commerce\u2019s Bureau of Industry and Security (BIS) conducts monthly spot checks on Anthropic\u2019s user data and risk-control records. Any discovery of Chinese users violating usage triggers warnings\u2014or fines.<\/p>\n
In late 2025, Anthropic was fined $12 million by BIS for \u201crisk-control vulnerabilities allowing some Chinese users unauthorized access to Claude Code\u201d\u2014a key reason it later intensified suspensions and implemented \u201cpreventive purges.\u201d<\/p>\n
By contrast, OpenAI and Google Gemini\u2019s \u201cgeographic restrictions\u201d are far more lenient\u2014not because they\u2019re \u201cfriendlier,\u201d but because their business models and strategic layouts afford greater \u201coperational space.\u201d<\/p>\n
OpenAI\u2019s deep Microsoft tie means Microsoft has extensive China operations\u2014requiring OpenAI to accommodate China-market needs, leading to relatively relaxed geographic risk controls, even tacitly permitting some Chinese users to access via third-party tools;<\/p>\n
Though Google Gemini complies with U.S. export controls, Google\u2019s limited China business and Gemini\u2019s core goal of scaling users mean it adopts \u201csee-no-evil\u201d tolerance toward Chinese users\u2019 violations\u2014rarely resorting to mass suspensions.<\/p>\n
Thus, Chinese users\u2019 suspension predicament is essentially a \u201ccasualty\u201d of U.S.-China AI decoupling.<\/p>\n
Anthropic\u2019s strict risk controls aren\u2019t merely its internal safety obsession, capital\u535a\u5f08, or factional infighting\u2014they\u2019re a direct manifestation of U.S. technology blockade policy. What we perceive as \u201cfalse positives\u201d is, to Anthropic, \u201cregulation evasion and policy violation.\u201d And what we view as \u201ctargeting\u201d is merely its self-preservation choice in great-power competition.<\/p>\n
Tripartite Balance: Safety Camp, Acceleration Camp, and Chinese AI Power<\/p>\n
Today, the global AI landscape is shifting from \u201cbipolar rivalry\u201d (U.S. vs. China) toward \u201ctripartite equilibrium\u201d\u2014anchored by Anthropic\u2019s Safety Camp, OpenAI\u2019s Acceleration Camp, and China\u2019s rapidly rising AI power\u2014interacting in competition and mutual restraint to shape AI\u2019s future trajectory.<\/p>\n
The safety-acceleration struggle continues escalating\u2014as analyzed earlier, no need to reiterate.<\/p>\n
While this U.S. internal conflict fuels Anthropic\u2019s strict risk controls, it also accelerates U.S. AI advancement: the safety camp deepens AI safety R&D; the acceleration camp pushes commercial and military AI applications. Their competition drives U.S. AI leadership\u2014still formidable.<\/p>\n
Now consider China\u2019s AI rise.<\/p>\n
Amid U.S.-China AI decoupling, China\u2019s domestic AI firms entered an \u201copportunity phase.\u201d<\/p>\n
Baidu\u2019s ERNIE Bot, Alibaba\u2019s Qwen, Huawei\u2019s Pangu, ByteDance\u2019s Doubao\u2014all iterate rapidly, narrowing technical and application gaps with U.S. AI.<\/p>\n
Especially in programming tools, domestic Chinese coding AIs (e.g., Doubao Code Assistant, ERNIE Bot Coding Edition) may lag top-tier models, but meet ordinary users\u2019 coding needs\u2014without geographic restrictions or harsh risk controls\u2014better fitting Chinese usage habits, gradually becoming \u201calternative choices\u201d for Chinese users.<\/p>\n
China\u2019s AI development follows a \u201cpragmatic, compliant, open\u201d path\u2014valuing both AI safety and commercialization, avoiding \u201czero-tolerance\u201d extremism or blind \u201cacceleration,\u201d seeking balance between safety and development.<\/p>\n
This path aligns with China\u2019s regulatory framework and ordinary users\u2019 needs\u2014gaining increasing market acceptance.<\/p>\n
Beyond this, Europe, Japan, and South Korea are also actively building AI industries\u2014seeking footholds in the global AI landscape.<\/p>\n
Europe emphasizes AI regulation, launching the EU AI Act to govern AI development while supporting domestic AI firms. Japan and South Korea boost AI R&D investment, focusing on AI applications in manufacturing, healthcare, and finance\u2014trying to catch up with the U.S. and China.<\/p>\n
Looking ahead, global AI competition will be a contest of \u201cideology, interests, and geopolitics.\u201d<\/p>\n
The safety camp seeks \u201ccontrollable development,\u201d the acceleration camp pursues \u201crapid ascent,\u201d and China\u2019s force aims for \u201cautonomous control and open cooperation.\u201d Their interplay will define AI\u2019s future\u2014and impact every ordinary person\u2019s life.<\/p>\n
Will Anthropic\u2019s Risk Controls Loosen\u2014or Stay High-Pressure?<\/p>\n
Finally, returning to the core question: Will Anthropic\u2019s suspension surge continue? Let\u2019s offer a forecast.<\/p>\n
First, one conclusion is clear: Anthropic\u2019s risk controls won\u2019t ease in the short term\u2014and may even tighten further.<\/p>\n
Three core reasons:<\/p>\n
First, Dario Amodei\u2019s safety conviction won\u2019t change easily. His \u201csafety religion\u201d is embedded in Anthropic\u2019s DNA. As long as he remains founder, this zero-tolerance logic won\u2019t shift.<\/p>\n
Second, the U.S. AI civil war and U.S.-China AI decoupling won\u2019t end soon. Accelerationist counterattacks and U.S. technology embargoes will continue pressuring Anthropic\u2014forcing it to maintain strict controls for compliance and self-protection.<\/p>\n
Third, Anthropic\u2019s business model doesn\u2019t require ordinary users. Its core focus is enterprise clients\u2014so ordinary-user attrition barely affects revenue. It has neither incentive nor need to relax risk controls for ordinary users.<\/p>\n
Long-term, however, Anthropic\u2019s risk controls may undergo \u201cdifferentiated adjustments.\u201d<\/p>\n
For instance, relaxing controls for ordinary users in compliant regions to reduce false positives; offering more flexible risk-control solutions for enterprise clients to meet diverse needs.<\/p>\n
Regarding Chinese users, I believe high-pressure restrictions will persist\u2014strictly limiting unauthorized usage. After all, complying with U.S. export control regulations is its survival bottom line.<\/p>\n","protected":false},"excerpt":{"rendered":"Deconstructing Anthropic\u2019s Descent into \u201cAccount-Banning Mania\u201d: This Is Not Merely the Founder\u2019s Obsession with Safety\u2014It\u2019s Also a Survival…\n","protected":false},"author":2,"featured_media":45596,"comment_status":"","ping_status":"","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[8],"tags":[53,3278,8581,8579,8577,8578,8101,8582,8576,8580],"class_list":["post-45595","post","type-post","status-publish","format-standard","has-post-thumbnail","category-anthropic","tag-anthropic","tag-blockchain","tag-blockchain-applications","tag-blockchain-in-depth-analysis","tag-blockchain-media","tag-blockchain-news","tag-blockchain-technology","tag-blockchain-trends","tag-techflow","tag-what-is-blockchain"],"_links":{"self":[{"href":"https:\/\/www.europesays.com\/ai\/wp-json\/wp\/v2\/posts\/45595","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/www.europesays.com\/ai\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/www.europesays.com\/ai\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/www.europesays.com\/ai\/wp-json\/wp\/v2\/users\/2"}],"replies":[{"embeddable":true,"href":"https:\/\/www.europesays.com\/ai\/wp-json\/wp\/v2\/comments?post=45595"}],"version-history":[{"count":0,"href":"https:\/\/www.europesays.com\/ai\/wp-json\/wp\/v2\/posts\/45595\/revisions"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/www.europesays.com\/ai\/wp-json\/wp\/v2\/media\/45596"}],"wp:attachment":[{"href":"https:\/\/www.europesays.com\/ai\/wp-json\/wp\/v2\/media?parent=45595"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.europesays.com\/ai\/wp-json\/wp\/v2\/categories?post=45595"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.europesays.com\/ai\/wp-json\/wp\/v2\/tags?post=45595"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}
![\"image\"](\"https:\/\/www.europesays.com\/ai\/wp-content\/uploads\/2026\/05\/20260520141150958246.png\")
<\/p>\n![\"\"](\"https:\/\/www.europesays.com\/ai\/wp-content\/uploads\/2026\/05\/20260520141150824649.jpeg\")
<\/p>\n![\"image\"](\"https:\/\/upload.techflowpost.com\/upload\/images\/20260520\/20260520141150801567.png\")
<\/p>\n![\"image\"](\"https:\/\/upload.techflowpost.com\/upload\/images\/20260520\/20260520141150914454.png\")
<\/p>\n![\"image\"](\"https:\/\/www.europesays.com\/ai\/wp-content\/uploads\/2026\/05\/20260520141150338376.png\")
<\/p>\n![\"image\"](\"https:\/\/www.europesays.com\/ai\/wp-content\/uploads\/2026\/05\/20260520141151415584.png\")
<\/p>\n