A former Anthropic employee says working at the AI company left him burned out and disillusioned. Ivan Vendrov was on the technical staff from late 2022 to mid-2023. He described his time there in a long Substack post, days after Anthropic filed to go public.
How to read the account
Vendrov now works at Midjourney, according to his LinkedIn. He opened by saying the post should not be taken as gospel. He wrote from memory, gave the gist of conversations rather than exact words, and called it the experience of one person in a junior role during a short window. Anthropic declined to comment. Vendrov did not respond to Business Insider.
Joining the company
In 2022, Vendrov was weighing contract offers from OpenAI and Anthropic. He chose Anthropic after lunch with a cofounder, who said the company was already talking with regulators in Washington. Then a recruiter asked if he would pledge two-thirds of his equity, meaning his share of company ownership, to charity. The recruiter said all the cofounders had done so. Vendrov was unsure whether the offer depended on it, so he said yes.
Anthropic's S-1 filing, the document a company files to sell shares to the public, describes a donation-matching program. Workers can pledge a "meaningful portion" of their equity awards. The company reported that 50% of its employees are included.
Life inside the office
- The office was cramped and ugly, but stocked with fresh croissants.
- A masseuse later came twice a week, but he says he could never book a slot.
- An FBI agent warned the company was likely a target of multiple foreign intelligence agencies.
- After a sushi chef visited, a cofounder wrote on Slack that it set the wrong tone.
By his recollection, the cofounder wrote, "We're not here to get rich and eat caviar," and floated future dinners of rice and beans instead.
Why he left
Vendrov wrote of sleepless nights, walks with other stressed-out colleagues, and moments when he felt he could not be honest about his emotions. Eight months after signing, he wrote a resignation letter. He considered sharing his doubts, but worried they would unsettle colleagues he saw as smarter, more competent and more solidly well-intentioned than himself. He instead called Anthropic an "unbelievably good organization."
Why it matters
Companies rarely let outsiders see how they work in their early days. This is a named employee describing a workplace where fear of AI disaster sat beside the push to build safe, commercial AI. It is one person's memory of a short period, and Anthropic has not responded to it.
The takeawayOne former employee's memory is not the whole story, but it offers a rare look inside Anthropic as it prepares to sell shares to the public.