PBC’s IPO filing offers a rare look into how seriously it takes threats to its executives’ and employees’ physical safety.

The company spent a combined $4.1 million on security for Chief Executive Officer Dario Amodei and President Daniela Amodei last year, which was about 30 times the roughly $135,600 median spent by S&P 500 companies, according to executive compensation research firm Equilar.

“Those values are larger than what we typically see companies pay for security,” said Courtney Yu, Equilar’s director of research. But some high-profile companies spend far more on protection for their top executives, especially after the 2024 assassination of the UnitedHealthcare CEO, he said.

AI leaders have already become targets. In April, San Francisco police arrested a man for throwing a Molotov cocktail at OpenAI CEO Sam Altman’s home.

Anthropic’s spending would rank seventh among S&P 500 companies, behind firms such as Meta Platforms Inc., which spends $22.5 million on security for CEO Mark Zuckerberg and his family, and Alphabet Inc., which spends $8.8 million for CEO Sundar Pichai.

San Francisco-based Anthropic spent $2.8 million on Dario’s security, placing him in the top 5% among S&P 500 executives with reported security spending, according to Equilar. Daniela’s security totaled $1.3 million, placing her in the top 10%.

Anthropic, the company behind the chatbot Claude, faces growing public backlash from those who fear artificial intelligence will take their jobs and those angered by the construction of sprawling data centers in their communities. “Heightened public attention on AI has increased physical and informational security risks,” the company said in its filing announcing the planned initial public offering.

Earlier this year, Bloomberg asked the Amodei siblings in an interview how the attack on Altman’s home was affecting them. “This is a time technologically, politically, where unfortunately there’s just a lot more rhetoric and words that I think can lead to bad outcomes and bad things happening,” Daniela said. “Obviously, it’s something we think about at Anthropic. We try our best to keep ourselves and our staff safe.”

“It was scary to me too,” Dario said. “I mean, this is like a less savory aspect of the exponential, right? That as AI gets more and more of a big thing, it becomes just such a big deal to society. There’s more attention on it. And that means that more people are paying attention and 0.1% of those people are crazy.”“We need to protect ourselves and the company,” he added. “We're doing everything we can.”Threats to employees may also make it harder to hire, the company said.

“We have experienced threats to employee safety and attempts to obtain confidential information through physical or digital means,” Anthropic said in its filing. That “could adversely affect employee morale, productivity, and our ability to attract, retain and motivate highly skilled personnel.”Those risks may be magnified by the company’s remote work policy, it said, which allows employees to work from home most of the time.

That disclosure is “fairly unique,” according to Rob Peters, a senior director at Intelligize, an SEC filings research platform. Few other public companies explicitly include risks to employees’ physical security in their filings, he said. One such example is Harmony Gold Mining Co., a South African company that identifies gold theft as a potential threat to employees.

Anthropic also noted that employees and contractors may be coerced or bribed to divulge sensitive information or access to internal systems by state-sponsored actors or competitors.

That concern is of particular note given the arrest of an OpenAI employee this month who allegedly planted a hidden recording device in an executive conference room at one of OpenAI’s San Francisco offices. The employee, a Saudi Arabian national, wasn’t authorized to access the room, which often hosted confidential meetings.

To address the threats to employee safety, Anthropic said it must hire “specialized security personnel” who may be difficult to recruit.

The company already has an around-the-clock security operation in place devoted to protecting workers, according to former staffers. All employees and contractors are provided with safety training and have access to a hotline if they’re in danger. One former employee, who asked not to be named, was surprised and impressed when the security team called to offer evacuation assistance immediately following a nearby shooting. Anthropic declined to comment. The filing also noted that bad actors have attempted to infiltrate the company via its hiring process.

“We have been and expect that we will continue to be targeted by individuals seeking employment under false identities, or with ulterior or malicious motives for seeking employment with us,” Anthropic said.

Companies across industries have seen a growing number of fake job candidates in recent years, whether individual scammers or state-sponsored hackers. Still, Anthropic’s disclosure is unusual: Companies typically describe the threat from bad actors in their filings in terms of cyberattacks, rather than in terms of the hiring process, Peters said.