Give Anthropic its due first. Setting aside 5% of a company worth hundreds of billions of dollars is more than most of its rivals have offered, and it comes from a CEO who has actually said out loud that AI could wipe out half of entry-level white-collar jobs and push unemployment into double digits. Most of Silicon Valley says the opposite in public and builds the same machine in private.
But look at what is being proposed. A company says its product may throw millions of people out of work. It then announces that it will decide, on its own, how much to set aside for those people, who counts as one of them, and what form the help takes. That is not a safety net. It is a private company writing its own damage settlement before the damage arrives.
The structure is the tell. Anthropic chose a wholly owned LLC over a tax-exempt nonprofit because it allows 'different forms of giving,' including 'direct distributions to specific groups.' Flexibility is the word of art here. A tax-exempt nonprofit generally has to file a public annual return with the IRS, and it is tied to a charitable purpose. An LLC owned by Anthropic answers to Anthropic, and nothing in the announcement promises comparable public reporting. The details on governance, valuation and structure are, we are told, coming 'in coming months.' So the public is asked to applaud first and read the terms later.
Then there is the sliding scale. The board 'could' add two percentage points for every 50% jump in the company's value above a baseline it won't disclose. Notice the word 'could,' and notice the premise. The fund grows only if Anthropic wins enormously. Workers get a larger slice only after investors have already been made far richer. If the AI boom stalls, the money shrinks just as people need it. If the boom delivers, the fund is still a rounding error beside the 95% held by investors, insiders and the giant corporations that bankrolled the company.
The pledge not to use fund equity for lobbying sounds principled until you ask who is excluded from the arrangement. The fund won't lobby. Anthropic can. So can the industry it belongs to, which is spending heavily to shape, soften and preempt every serious attempt to regulate it. A charity that can't lobby, attached to a company that does, is a neat division of labor. One arm gets the goodwill and the other gets the policy outcomes.
The Left should say plainly what this is and what it isn't. It's better than nothing, and nothing has been the industry norm. But it is a gift where a right is needed. Displaced workers shouldn't depend on the continuing generosity of a board that retains discretion over every dollar. The models were built on the accumulated output of millions of writers, coders and artists, many of whom were never asked and never paid. Anthropic itself agreed to a $1.5 billion settlement with authors over pirated books. That should settle the question of whether this wealth came from nowhere.
The actual answers are public and unglamorous: wage insurance, serious retraining money that isn't a punchline, stronger bargaining rights for workers whose employers deploy these systems, and a tax code that captures gains from automation instead of letting them pool in a few cap tables. Those require democratic decisions, which means they require political fights, which companies like Anthropic would rather be seen as having pre-empted.
There's a danger in being too grateful. Every voluntary pledge from a powerful company becomes an argument against mandatory policy. 'We're already handling it.' Legislators will hear that line, and some will be glad to.
Anthropic can fund whatever it likes. The test is whether it will support the binding rules that don't depend on its goodwill, publish every term of this fund, and accept independent control over where the money goes. Until then, this is a promise from the party that holds all the power, about how kindly it intends to use it.
