A PERSONAL GRIEVANCE. NOW IN 16 BITS.
MISANTHROPIC✳
The morally superior way
to tell me I’ve hit my limit.
A $200 space opera about the company that sells you the future, warns you about the future, and asks you to come back later to use the future.
BY AXL IBIZA • GALAGA-INSPIRED SATIRE • SEPTEMBER 2026
“The future of humanity is at stake.
Your five-hour window has expired.”
01 / WHY I’M STILL MAD
A note from the person
holding the joystick.
I came for a tool.
I got a chaperone.
What I hate is the distance between Anthropic’s moral self-image and the experience of being its customer.
The pitch is unusually grand. These are the careful people. The thoughtful people. The people who understand the danger and can be trusted to guide everyone else. In its February 2026 statement on advertising, Anthropic says it wants Claude to act “unambiguously in our users’ interests.” That is the standard I’m holding it to.
Then I look at the product history: extra ceilings, confirmed quality failures, restrictions tied to the hour of the day, changes to where subscription usage can be spent, and announcements that have to be reconciled with reversals. Each incident has its own explanation. Together they become the relationship.
My own irritation has an equally mundane core: read the material, understand what I asked, and do the work. In my saved March 2026 sessions, I was already correcting an assistant for responding before reading an attached spreadsheet, and separately asking for a custom skill’s patronizing tone to be removed. Those are specific interactions, including custom behavior. They explain why the polished institutional voice does so little for me. Private archive notes ↓
A company doesn’t get to turn “we care more” into a substitute for treating people better.
That is where the moral branding becomes aggravating. A routine product failure arrives wrapped in an exceptional self-conception. I’m allowed to judge that combination harshly. I’m allowed to expect more from a company that asks to be trusted with more.
I build AI systems. I want people to have useful, capable tools. Being pro-AI does not require a loyalty oath to the company that makes the tool, and criticism of Anthropic does not require pretending every rival is wonderful. This is my case against this company’s bargain.
THE ENEMIES ARRIVE IN FORMATION
Five-hour limits.
Weekly limits.
Unlimited explanations.
How Anthropic describes the shared usage limits ↗02 / THE LONG DESCENT
The price is concrete.
The bargain moves.
Follow the public record from the Max launch
through the policy reversals. The source links
stay beside the claims. So does the context.
Insert $200. Receive “Max.”
Anthropic introduced two Max tiers: $100 for 5× Pro usage and $200 for 20×, with priority access to new capabilities. The pitch was more room for people who use Claude intensively.
Claude · Max plan announcement ↗A premium plan for power users. Lovely. Remember the words “power users.” They become relevant when using the product enthusiastically starts being described as a capacity problem.
“Thank you for being our most dedicated customer. Please be less dedicated.”
OPEN FLIGHT LOG What happened next + context
Max was sold with usage limits. The criticism is the changing bargain, not a claim that Anthropic promised unlimited use.
Your five-hour wall now has a weekly wall.
Anthropic announced additional weekly subscription limits for late August, saying fewer than 5% of users would be affected under then-current usage patterns. It cited continuous use and unusually expensive consumption.
Anthropic · subscription limits announcement ↗Sell the higher tier to the people doing the most. Introduce another ceiling for the people doing the most. Explain that very few people are affected, as if being a small percentage of customers makes your paid work less interrupted.
“Only a small percentage of pilots will hit this asteroid.”
OPEN FLIGHT LOG What happened next + context
This was an announced policy change. “Less than 5%” was Anthropic’s estimate, not an independently measured outcome.
It wasn’t all in your prompt.
Anthropic acknowledged three infrastructure bugs that intermittently degraded responses in August and early September. Users had reported problems in early August; an investigation followed later that month. About 30% of active Claude Code users in the routing-error period had at least one message sent to the wrong server type.
Anthropic · three infrastructure issues ↗The customer gets to wonder whether they forgot how to ask a question. Eventually the company produces a technical explanation. The lost work happened before the explanation arrived. A postmortem is useful. It is a peculiar premium benefit.
“Your lived experience has successfully passed our evaluation suite.”
OPEN FLIGHT LOG What happened next + context
Anthropic reported fixes and explicitly denied deliberately lowering quality with demand. These bugs do not establish intentional degradation.
Maxed out. Still subscribed.
A public issue collected complaints across Pro and both Max tiers about weekly allowances disappearing in one or two days. Contributors described being unable to continue the work they bought the subscription to do.
Public Claude Code issue #9424 ↗There is a special kind of customer experience in which the payment recurs monthly, the permission to work resets weekly, and the explanation requires a community megathread.
“Your subscription is active. Your productivity is on cooldown.”
OPEN FLIGHT LOG What happened next + context
These are customer reports, not an audit of every subscriber’s allowance or proof of a universal reduction.
Productivity. Terms and time zones apply.
Anthropic announced that Free, Pro, and Max users would consume their five-hour session allowances faster during weekday peak hours. Weekly limits were unchanged. The announced window was 5–11 a.m. Pacific.
Anthropic · session limits update ↗An assistant for your workday, with less room during the workday. The calendar has joined the enemy fleet. I would like to complete the task without first consulting the electricity tariff for a different continent.
“Have you considered having your deadline at a less popular time?”
OPEN FLIGHT LOG What happened next + context
On May 6, Anthropic announced the removal of this peak-hour reduction and doubled Claude Code’s five-hour limits.
Anthropic’s May 6 reversal ↗Same model. Different cockpit. Pay again.
The April change moved third-party harnesses such as OpenClaw off included subscription usage and onto separate billing. A one-time transition credit and refund route were reported.
TechRadar · third-party harness billing ↗For people who build their own agent workflows, the interface is part of the work. Restricting where included usage follows the customer makes “your AI” feel very much like somebody else’s tollbooth.
“You may explore the universe. Please remain inside our cockpit.”
OPEN FLIGHT LOG What happened next + context
This describes the April event. Later Agent SDK policy proposals and their pause are a separate development, covered next.
The new rule has entered a holding pattern.
An announced switch would have moved Agent SDK and claude -p usage onto a separate monthly credit. The help page now begins with a June 15 update: the change is paused, the proposed credit is unavailable, and those uses still draw from subscription limits.
Claude Help Center · Agent SDK plan update ↗The developer’s side quest: reconcile the announcement, the documentation, the effective date, and the reversal before deciding whether the thing that worked yesterday is still included today.
“We have updated the update. Please update your understanding.”
OPEN FLIGHT LOG What happened next + context
The proposed June 15 switch did not take effect as announced. This page records the pause, rather than presenting the abandoned plan as current policy.
03 / THE SIDE QUESTS
Even the fine print
has reinforcements.
Your work.
Their next model.
The consumer-policy update invited Free, Pro, and Max customers to allow chats and coding sessions to be used for training, with five-year retention if they opted in. Commercial services were treated separately. Anthropic said users could decline and change their preference.
I’m paying for a tool. Now I also need to inspect whether my work is becoming an input to its next generation. The choice exists. So does the new administrative burden.
Anthropic’s announcement and opt-out explanation ↗To complain about AI,
first talk to AI.
Anthropic’s support documentation directs users through Fin, its AI support bot. Pro and Max subscribers can seek further assistance from Product Support; the page describes an escalation route.
An AI standing between an unhappy AI customer and the AI company is almost too tidy a joke. Human escalation matters. So does how much effort it takes to get there.
Anthropic’s support process ↗One Max subscriber’s public account · anecdotal ↗04 / DARIO FM · ALWAYS TRANSMITTING
Sell the future.
Scare the present.
The recurring character has a real public voice.
Here are the words behind the mothership.
FM 20× ON AIRCAPTAIN
CONTRADICTION
“Please fear the technology.
Please subscribe to the technology.”
The utopia trailer
Machines of Loving Grace
Amodei’s optimistic essay describes enormous potential benefits across health, science, prosperity, and human life. It matters because his public position includes enthusiasm as well as warnings.
Dario’s essay ↗Keep this trailer playing on the screen next to the disaster trailer. The contrast is the character.
Your career: approaching enemy fire
Half of entry-level white-collar jobs
Amodei warned of losing half of entry-level white-collar jobs and 10–20% unemployment within one to five years.
Axios interview · prediction, not an observed result ↗My objection: this is an extraordinarily frightening message for people trying to enter a profession. When an AI CEO supplies that headline, public anxiety about AI should not surprise him.
The civilization boss fight
The Adolescence of Technology
His follow-up essay explores loss of control, destructive misuse, authoritarian power, economic disruption, and indirect consequences. He also argues for responses to those risks and criticizes sensationalism.
Dario’s essay ↗I read this combination as fear-driven positioning: sell the extraordinary power, narrate the catastrophe, and occupy the role of responsible guide. That is my criticism of the rhetoric, not proof of a secret motive or a measured effect on public opinion.
I think this messaging helps make AI sound frightening to the people who might otherwise use it. That is an editorial judgment about the public rhetoric. The sources establish what he said; they do not establish that he deliberately incited hostility, or quantify how much public opinion he changed.
05 / THE HALO HAS PATCH NOTES
Apparently, principles
can be refactored.
This is where the “morally superior alternative” positioning really loses me.
In February 2026, Anthropic revised its Responsible Scaling Policy. Its own explanation separates unilateral plans from industry-wide recommendations and describes Frontier Safety Roadmap goals as nonbinding public targets.
Anthropic · RSP version 3.0, February 24, 2026 ↗TIME reported the removal of the earlier central pause pledge, quoting chief science officer Jared Kaplan’s explanation that stopping training would not help while the wider industry continued.
TIME · interview with Anthropic leadership ↗My objection is the asymmetry. Customers are expected to adapt to the company’s constraints. The company reserves room to adapt its own commitments to competitive reality. Then the brand still gets to wear the halo.
The revision did not abolish all safeguards. It introduced roadmaps, risk reports, and review mechanisms. The criticism concerns the changed commitment and the branding around it. Read the policy and subsequent updates ↗
06 / BONUS STAGE · DANIELA JOINS THE FLEET
Same galaxy.
Executive polish.
Daniela arrives with a golden ticket and the executive composure this space opera desperately lacks. Anthropic identifies her as its co-founder and president. Here, she gets a fictional executive cameo in the family mothership.
Anthropic’s leadership page ↗“We’ve listened to your feedback.
It now has its own five-hour limit.”
The joke is invented. It does not attribute the individual policy decisions above to her.
07 / MY VERDICT
MORAL SUPERIORITY
SOLD SEPARATELY.
The halo makes the ordinary failures worse.
I expect a technology company to have constraints. I object when exceptional moral authority becomes the packaging for a bargain this unstable. In my judgment, that combination makes Anthropic worse than the image it sells: more grandiose about its responsibility, less convincing in the everyday relationship.
I don’t owe a subscription my allegiance. The company owes its customers a product worth paying for. Useful AI. Clear terms. Respect for my agency. A stable understanding of what I bought. If your brand is responsibility, start there.
PRIVATE ARCHIVE NOTES & EDITORIAL METHOD
The personal passage refers to March 24 and March 26, 2026 Claude-session records: a request to read an attached spreadsheet before responding, and a request to remove a patronizing tone from the custom ritual-affirmation skill. These are paraphrases, not public transcript quotations. The latter concerns a custom skill and should not be generalized to Claude’s default persona. The underlying files contain private material and are not published.
This page uses dated public announcements, an official postmortem, direct essays, reporting, and explicitly identified customer accounts. It does not treat social-media complaints as measured prevalence. Source links accompany the factual passages; historical restrictions include their subsequent reversals. It is a selected, documented account, not a claim to enumerate every incident or every private session.
MISANTHROPIC / AN ORIGINAL SATIRE BY AXL IBIZA
Galaga-inspired visual storytelling. Original caricatures created with AI at Axl’s direction. No affiliation with Anthropic or the creators of Galaga.
Public record checked September 6, 2026. Commentary is opinion. Character dialogue is invented. Policy pages can change; dated events refer to the version described. Private documents are not reproduced.