• Asana CEO Dustin Moskovitz, who made billions from co-founding Facebook, has long been concerned about the risks of runaway artificial intelligence.
• Moskovitz has been snapping up shares of Asana and is now majority owner, with plans to give away the majority of his wealth to charities.
• He was an early backer of OpenAI and counts "potential risks from advanced AI" as one of the focus areas of his philanthropy.

The typical playbook for a successful tech founder looks something like this.

Start a company with full ownership. Sell off significant chunks to venture investors as the business progresses. Eventually become a minority owner. Take the company public. Sell more stock over time.

Asana's Dustin Moskovitz took that playbook and completely rewrote the ending.

Moskovitz, who is still known by many as a co-founder of Facebook, started Asana in 2008 to make work more collaborative through software. By the time he took the company public through a direct listing in 2020, his ownership stood at about 36%.

Then, he went on a buying spree. Following the purchase of 480,000 Asana shares in June, Moskovitz's ownership swelled to 111.4 million shares, representing over 51% of outstanding stock. In March, Asana disclosed that Moskovitz had a trading plan to buy up to 30 million more of its Class A shares this year, sending the stock up almost 19% the next day.

"It's been a wild two years in the market and there have been some interesting buying opportunities," Moskovitz said in an interview with CNBC.

Even after rallying 66% this year, Asana shares are more than 80% below their record high from late 2021.

For Moskovitz, who has a net worth over $12 billion — mostly from his early stake in Facebook, now Meta — becoming majority owner of Asana isn't about control. Rather, he sees it as the best way to invest to support his philanthropy.

In 2010, Moskovitz signed the Giving Pledge, a promise by some of the wealthiest people in the world to donate most of their fortunes to charity. Moskovitz and his wife, former journalist Cari Tuna, dole out their funds through Good Ventures, based on recommendations from Open Philanthropy.

When it comes to spending that money, there's no greater concern to Moskovitz than the future of artificial intelligence.

Good Ventures donated $30 million to startup OpenAI over a three-year period in 2017, long before generative AI or ChatGPT had entered the public lexicon. OpenAI, which is now worth about $30 billion, was started as a nonprofit, and Open Philanthropy said at the time it wanted "to help play a role in OpenAI's approach to safety and governance issues."

One of the 10 focus areas Open Philanthropy lists on its website is "potential risks from advanced AI." The organization recommended a $5 million grant to the National Science Foundation to back research on methods of guaranteeing the safety of artificial intelligence systems, and $5.56 million to the University of California at Berkeley for "the creation of an academic center focused on AI safety." In total, Open Philanthropy says it's given over $300 million in the focus area through more than 170 grants.

"I definitely think there's a big risk there — something I spend a lot of time thinking about," Moskovitz said.

Moskovitz co-founded Facebook with Mark Zuckerberg, Chris Hughes and Eduardo Saverin at Harvard University in 2004. He became a billionaire after Facebook's 2012 initial public offering, holding more shares than any individual other than Zuckerberg.

Even after snapping up additional Asana shares in 2022 and 2023, his ownership sits at about $2.6 billion, less than the $4.6 billion in Facebook stock he owns, according to FactSet.

"I'm just in a unique position, where I came to the table with an existing source of wealth," Moskovitz said. "So even things that look like gigantic purchases, it's still a relatively normal sort of portion of my net worth relative to other founders."

Moskovitz has agreed not to buy all outstanding Asana shares or even acquire ownership of 90% of the common stock. He will also keep a majority of its directors independent, in compliance with the rules of the New York Stock Exchange, according to a filing.

Moskovitz declined to talk about whether he was buying up shares to prevent activist investors from coming in and trying to force change. Activists have been busy in the cloud software space, most notably at Salesforce, which responded to pressure by expanding its buyback program and bolstering profits.

Recently, Moskovitz's worlds collided.

OpenAI vaulted from niche startup to the hottest thing in tech after releasing ChatGPT in November. Before that, Moskovitz was playing around with the company's DALL-E technology for converting text into images. He said OpenAI CEO Sam Altman set him up with a "labs account" in April of last year.

Following the ChatGPT launch, Moskovitz had some fun asking the chatbot to come up with objectives to help deal with California's housing problem.

Meanwhile, Asana joined the parade of companies that announced enhancements to their products with generative AI features that could take human input and present text, images or audio in response. Earlier this month, Asana said it had given some clients access to several generative AI features powered by OpenAI's models.

"Chat is just one paradigm for how you use these technologies," Moskovitz told CNBC. "When you're integrating them into workflows like work management, doing things like optimizing automation workflows or helping to make decisions — you can literally ask questions of the system and it'll give you a summary and a recommendation."

Moskovitz said more complicated tasks, such as adding structure to projects, is where "it really sorts of takes off in potential." Rather than just asking for specific answers, he said the power is in the technology to take "a bunch of information and sort of a vague goal" and then "give you something approximately in the right direction."

Asana could spend $5 million or more on OpenAI's technology next year, Moskovitz said, adding he was "very impressed by GPT-3," the company's prior large language model, "and was even more impressed by GPT-4," which was announced in March.

Moskovitz took six minutes out of Asana's 51-minute earnings call in early June to tout the company's approach to AI. He used the acronym 41 times, compared with 32 AI references by Microsoft CEO Satya Nadella on his company's earnings call in April. Microsoft is OpenAI's lead investor.

Asana is "just personally deeply connected to the AI labs that are leading the way," Moskovitz said.

The links are, in fact, quite deep. Altman invested in Asana in 2016. On Asana's earnings call, Moskovitz reminded analysts that his company and OpenAI "share a board member in Adam D'Angelo," a former Facebook technology chief who later started online Q-and-A startup Quora.

Moskovitz invested in AI startup Anthropic in 2021, the same year he co-invested with Altman in nuclear fusion startup Helion.

Similar to Altman, Moskovitz is also deeply bullish on AI and worried about the damage it can cause.

Moskovitz was one of many entrepreneurs who signed a statement in May, saying that "mitigating the risk of extinction from AI should be a global priority alongside other societal-scale risks such as pandemics and nuclear war." The missive came from the nonprofit Center for AI Safety.

But Moskovitz wasn't among the signatories of the nonprofit Future of Life Institute's open letter in March that called on AI labs to press pause on training the most sophisticated AI models for six months or more. Near the top of that list of signees was Tesla CEO Elon Musk, an early backer of OpenAI who has warned we should be very concerned about advanced AI, calling it "a bigger risk to society than cars or planes or medicine."

Moskovitz said Musk's fears aren't completely overblown and that they both want "to bring this technology into the world in a safe way."

"Elon kind of comes at it from multiple angles," he said. "I think we sort of share the view about potential existential risk issues, and maybe don't share the view as much about AI censorship and wokeism and stuff like that."

In December, Musk tweeted that "the danger of training AI to be woke — in other words, lie — is deadly."

Moskovitz has helped craft a 12-point list of possible policy changes for U.S. lawmakers to consider.

"The thing I'm most interested in is making sure that state-of-the-art later generations, like GPT-5, GPT-6, get run through safety evaluations before being released into the world," he said. "I think that will require regulation to coordinate all the players."

He even made up a word, in a tweet last month, to express his convoluted views.

"Excito-nervous for AI!" he wrote.

Correction: This story has been updated to remove an incorrect reference to the founders of Anthropic.

WATCH: Elon Musk creates A.I. startup called X.AI to take on OpenAI's ChatGPT