Anthropic Gets Its P45: The London Prat on Silicon Valley's Most Overdue Redundancy Notice

By The London Prat Technology and Corporate Accountability Desk

There are many ways to lose your standing in the technology sector. You can be disrupted, which sounds like a natural phenomenon but is in fact what happens when someone with more funding builds the same thing cheaper. You can be pivoted away from, which is the corporate equivalent of a partner announcing they have "decided to explore other options." You can be sunsetted, a euphemism so achingly passive-aggressive that whoever invented it deserves some kind of prize for creative obfuscation. Or you can simply receive your P45 — that most British of employment documents — handed over with a cup of tea, a complete absence of eye contact, and the clear understanding that the conversation is now over.

The London Prat brings you the full, unvarnished, satirically enhanced account of Anthropic receiving its P45. We have examined the evidence, interviewed our imagination, and produced what we believe to be the definitive analysis of events at one of Silicon Valley's most earnest artificial intelligence companies.

What Is Anthropic and Why Should Anyone Care

Anthropic is the San Francisco artificial intelligence company founded in 2021 by Dario Amodei, Daniela Amodei, and a group of researchers who had previously worked at OpenAI and left, as the polite version of the story goes, due to "differences in vision." The less polite version involves various accounts of internal tensions, strategic disagreements, and the kind of organisational drama that AI companies generate with the same reliable productivity with which they generate press releases about changing the world.

The company develops Claude — an AI assistant that is, depending on your perspective and your previous experiences with it, either a remarkable technological achievement representing a genuine step towards beneficial artificial intelligence, or the reason your colleague now submits reports that are grammatically perfect, comprehensively wrong, and mysteriously devoid of any actual knowledge of the subject they purport to address. Claude is, Anthropic is at pains to emphasise, safe. The company was founded on the principle that AI safety should be central to AI development, a mission that has attracted billions of dollars in funding from investors who are either genuinely committed to the principle or have calculated that "safety-focused AI" is a compelling regulatory arbitrage strategy. Possibly both.

The P45: Britain's Most Honest Document

For readers unfamiliar with the British employment system, a brief explanation. The P45 is the tax form issued by an employer to an employee upon termination of employment. Her Majesty's Revenue and Customs — now His Majesty's Revenue and Customs, though the form has not noticeably improved — requires employers to complete this document whenever someone stops working for them, for any reason. The form records your earnings to date, your tax paid to date, and your tax code. It is, in essence, the government saying: you are no longer employed here, here is the paperwork, please don't cause a scene in the car park.

The P45 has a cultural resonance in Britain that extends well beyond its administrative function. To give someone their P45 is to sack them. To receive your P45 is to have been sacked. The phrase carries with it the full weight of the British employment experience — the slightly too-formal handshake, the cardboard box of personal belongings, the farewell card signed by people who were never entirely sure of your surname. It is an honest document in a world of dishonest euphemisms, and The London Prat respects it accordingly.

The AI Safety Industry: Profitable Anxiety

One of the more remarkable features of the artificial intelligence industry in the 2020s has been the emergence and rapid growth of what might be called the AI safety industrial complex. Companies like Anthropic, research organisations like the Machine Intelligence Research Institute, and a rotating cast of academics, ethicists, and former Google employees have built substantial careers and organisations around the proposition that AI is extremely dangerous and that the people best placed to manage that danger are, coincidentally, the people currently building it.

The Financial Times has noted the particular economic logic of this arrangement: the more alarming your public statements about AI risk, the more essential your safety-focused company becomes, and the more readily investors open their chequebooks to ensure that the scary AI is being built by you, carefully, rather than by someone else, recklessly. It is a business model of considerable elegance, combining genuine intellectual substance with an incentive structure that rewards emphasising the dangers of the thing you are simultaneously building and selling.

Anthropic has raised billions of dollars, attracted partnerships with Amazon Web Services and Google Cloud, and positioned itself as the responsible adult in a room full of people racing towards capabilities that might, on current trajectories, produce systems of world-altering consequence within years rather than decades. Whether this positioning reflects genuine differentiation in approach or sophisticated marketing is a question that The London Prat finds genuinely interesting, without pretending to have a definitive answer.

What the P45 Represents in Practice

The specific circumstances of Anthropic's P45 — as explored in our full satirical account — involve the kind of corporate and competitive pressures that make the technology sector such a reliable source of entertainment for publications like ours. The AI industry has, since the public release of ChatGPT in late 2022, undergone a transformation of almost violent speed. Companies that were research organisations have become commercial enterprises. Research organisations that were commercial enterprises have become arms of major technology conglomerates. And the competitive landscape has shifted sufficiently rapidly that positions of apparent strength have a tendency to become positions of awkwardness in the space of a single product cycle.

OpenAI — from which Anthropic's founders departed — has continued its own extraordinary trajectory. Google DeepMind has consolidated and accelerated. Meta has released open-source models that have disrupted the economics of proprietary AI in ways that continue to reverberate. The competitive dynamics of a sector this young and this well-funded are, to use the technical term, absolutely mental, and the companies that navigate them successfully are those with the clearest sense of their actual competitive advantage, as distinct from their press release competitive advantage.

Dario Amodei and the Burden of Prophecy

A significant portion of Anthropic's public profile rests on the shoulders of its CEO, Dario Amodei, who has become one of the most prominent and articulate voices in public discourse about AI risk. Amodei's essays and interviews on the potential dangers and benefits of advanced AI are genuinely thoughtful, considerably more nuanced than most public commentary on the subject, and read by an audience that extends well beyond the technology industry into government, academia, and international policy.

The burden of Amodei's position is that he is simultaneously the prophet of AI danger and the salesman of AI products. His public statements about the potentially transformative — and potentially catastrophic — consequences of advanced AI necessarily coexist with quarterly business targets, investor expectations, and the commercial reality that Claude needs to generate sufficient revenue to justify the billions invested in developing it. The London Prat does not suggest that this combination is hypocritical — it is, rather, the inescapable condition of anyone who has decided that the best response to the risks of powerful AI is to build it more carefully than the competition, rather than not build it at all.

The London Prat's Considered Assessment

The London Prat has covered technology companies with a mixture of fascination and scepticism since we first noticed that the industry's capacity for self-congratulation was exceeded only by its capacity for generating stories of spectacular corporate hubris. Silicon Valley's greatest gift to satirists is its absolute conviction that every new product is not merely a business but a civilisational project, not merely a software release but a step towards human flourishing, not merely a quarterly earnings event but a moment in the long arc of history bending towards justice.

Anthropic, to be fair, is more self-aware than most. The company's founders have consistently acknowledged the genuine risks of what they are building whilst arguing that the risks of not building it carefully exceed the risks of building it at all. This is a coherent position, and The London Prat does not mock it. We do, however, reserve the right to note that when a company with this level of funding, this level of press coverage, and this level of civilisational ambition encounters the ordinary turbulence of commercial competition and market pressure, the gap between the grand narrative and the prosaic reality is precisely the gap in which satire lives.

The P45 arrives for everyone eventually, even if the accompanying paperwork runs to several billion dollars. Read our full satirical account here.

Auf Wiedersehen, amigo!

Anthropic is an AI safety company founded in 2021 by Dario Amodei, Daniela Amodei, and other former OpenAI researchers. It develops the Claude family of AI assistants and has raised significant venture capital and strategic investment from Amazon and Google. A P45 is the British tax and employment document issued to workers upon leaving a job; in common usage, "getting your P45" means being sacked. The company has positioned itself as a safety-focused alternative to other AI developers, arguing that advanced AI must be developed carefully to avoid catastrophic outcomes.