The 18-month lie: AI executives keep promising automation deadlines they keep missing

Review

In February 2026, Microsoft AI CEO Mustafa Suleyman told the Financial Times that white-collar work would be “fully automated by an AI within the next 12 to 18 months.” It was a striking claim. It was also, almost word for word, a claim the AI industry had already made and missed several times over. The remarkable thing is not that he said it. The remarkable thing is that saying it cost him nothing.

The ledger of missed deadlines

An accountability ledger compiled by Sloppish documents the pattern:

• Elon Musk (March 2024): “AI will probably be smarter than any single human next year.” Deadline: end of 2025. Status: MISSED. At Davos in January 2026, Musk told Larry Fink that “we might have AI that is smarter than any human by the end of this year, or no later than next year” — the same claim, slid forward by a year, with no acknowledgment that the previous one had failed.

• Sam Altman (January 2025): “In 2025, we may see the first AI agents ‘join the workforce’ and materially change the output of companies.” Deadline: end of 2025. Status: UNMET. The hedges — “we may see,” “first” — are engineered to be technically unfalsifiable while still functioning as a headline.

• Dario Amodei (March 2025): “I think we’ll be there in three to six months, where AI is writing 90 percent of the code. And then in twelve months, we may be in a world where AI is writing essentially all of the code.” Deadline: September 2025 / March 2026. Status: MISSED. The most generous public figures put AI-generated code at major tech companies at 25-30%. Even at Anthropic itself, the 90% claim does not survive scrutiny.

• Mustafa Suleyman (February 2026): Most white-collar tasks “fully automated” by AI. Deadline: 2027-2028. Status: PENDING.

The accountability gap

What happened to Amodei and Anthropic after the 90% code deadline lapsed? The company’s run-rate revenue climbed from roughly $1 billion to over $5 billion by August 2025. In February 2026, it closed a $30 billion Series G at a $380 billion post-money valuation. The missed prediction did not dent the company. The company got several times more valuable while the prediction sat there, unmet, in the public record.

The deadline passed. The valuation went up. That is the entire incentive structure in one sentence. When being wrong is free, the rational move is to keep being wrong, loudly, because the upside — funding, recruiting, narrative control, share price — accrues whether or not the prediction lands. The prediction is the product. The deadline is marketing copy with an expiration date nobody checks.

Why this is propaganda

These are not good-faith estimates that turned out to be off. They are dated, specific, measurable claims made by people who control billions in capital and whose words move markets. They are repeated by journalists who rarely circle back to check. And they are never retracted. The pattern is identical across every major AI company: make the prediction, collect the valuation, extend the deadline, repeat. That is not science. It is marketing dressed in the language of forecasting, and it functions as propaganda because it shapes policy, investment, and public expectation without any accountability for accuracy.

Verdict: False. The predictions were wrong. The predictors were rewarded. The mechanism is the absence of consequence.