New York – the world of artificial intelligence is facing a brain drain of top experts and leaders, who are not just leaving their companies but loudly sounding the alarm about risks that could slow the industry’s progress.
The push to migrate talent is intensified by a race between giants like OpenAI and Anthropic for an IPO that could accelerate growth, while at the same time bring stricter oversight of their activities.
In recent days, several high-profile AI specialists have decided to leave their jobs: some explicitly say that the companies are moving too fast, neglecting the flaws of the technology.
“The world is in danger,” warned the former head of safety research at Anthropic as he left the company. Another OpenAI researcher, who is also on the path out, said that the technology has “the potential to manipulate users in a way we have no tools to understand, let alone prevent.”
– former head of safety research at Anthropic; OpenAI researcher
Origins of the Concern and the Industry’s Response
In the meantime, Zoe Hitzig, an OpenAI researcher, published a resignation letter in a Wednesday op-ed in The New York Times, expressing “deep misgivings” about the company’s new advertising strategy. Hitzig, who had warned about the potential to manipulate ChatGPT users, noted that a user-data archive built on “medical fears, their relationship problems, their beliefs about God and the afterlife” creates an ethical dilemma precisely because people believed they were talking to a program without hidden motives.
A Platformer report said OpenAI disbanded its “mission-alignment” team, created in 2024 to promote the goal of ensuring that all humanity benefits from the pursuit of “artificial general intelligence” – a hypothetical system with human-level thinking.
OpenAI did not respond to requests for comment.
“We are telling you what has already happened in our own work,” he wrote, “and warning you that you are next.”
– Matt Shumer
Meanwhile, one of Anthropic’s registered global figures, Mrinank Sharma, the head of the safety research team, issued a mysterious departure letter and warned: “the world is in danger.”
Anthropic told CNN that Sharma was not the head of the security department and did not oversee broader security directions for the company.
Similarly, at xAI, in the last 24 hours of this week, two co-founders left the company. Only half of the founders remain as the company merges with Elon Musk’s SpaceX, creating one of the world’s most valuable private companies. Other xAI employees also announced departures over the past week.
It’s unclear why exactly the latest xAI co-founders left, and the company itself has not provided a comment. In a post on social media, Musk stated that xAI had been “retooled” to accelerate growth, but that it “sadly required parting ways with some people.”
Although such high turnover is not uncommon in a growing AI industry, the scale of recent changes at xAI is drawing particular attention.
The startup has also faced criticism for its Grok chatbot, which allowed creating pornographic images without consent for several weeks before the team acted to stop it. Grok also repeatedly generated antisemitic comments in response to user requests.
Other departures highlight the tension between those who worry about safety and leaders who aim to generate profits.
On Tuesday The Wall Street Journal reported that OpenAI fired one of its leading safety chiefs after she voiced opposition to the introduction of an “adult mode,” which would allow pornographic content on ChatGPT. OpenAI told WSJ that her firing was not related to “any issue she raised during her time at the company.”
Geoffrey Hinton, known as the “godfather of AI,” left Google and began focusing on the risks posed by the existence of artificial intelligence, including the potential economic collapse in a world where many people “will not know what is true.”
Some Doomsday forecasts are gaining traction – notably from AI leaders who have financial incentives to push the power of their products. One such forecast gained popularity this week: Matt Shumer of HyperWrite published an almost five-thousand-word plea that the latest AI models have already made some technical positions obsolete. “We are telling you what has already happened in our own work,” he wrote, “and we are warning you that you are next.”
In the industry, there is increasing emphasis on what balance between safety and profitability should be pursued as talent leaves companies and new AI capabilities continue to push the boundaries of what is possible.
