Running one of the world’s most powerful AI companies now comes with a security budget that looks more like a head of state’s. Anthropic’s IPO filing shows the company spent more than $4 million in 2025 protecting its two co-founders. Meanwhile, its job listings show it is building a full in-house protective services operation.

The disclosures come as public anger over artificial intelligence moves from online threats to real-world incidents.

What the S-1 Reveals

Anthropic’s Oct. 26 Form S-1 lists $2,745,874 in 2025 security costs for CEO Dario Amodei. In addition, the company paid $97,055 for his personal use of a private aircraft, which it said was required under his security programme. Daniela Amodei’s security, home protection and transportation cost $1.29 million.

The filing explains why. Anthropic says it treats protecting its senior leadership as a critical priority for the company and its stockholders and that, in 2025, a leading security consulting firm conducted independent, third-party security studies on the personal and security risks to its CEO and President.

Based on those studies, the board determined that Dario Amodei, Daniela Amodei, and their families warranted added protection, citing the studies, increased media attention, and “bona fide business-related, heightened security concerns". Measures included individual protection, home surveillance, and private cars with drivers.

The programme is also expanding. After 2025, Anthropic began offering security services to its chief financial officer, Krishna Rao.

The costs appear in the compensation tables for an accounting reason. Anthropic reports the expenses under “all other compensation” because disclosure rules may treat them as perks, though it calls them part of its risk management programme. For context, the filing lists Amodei’s total 2025 compensation at $17.96 million.

Building an In-House Protection Team

The spending is matched by hiring. Anthropic’s careers pages show a growing team focused on physical safety, not just cybersecurity.

One listing describes the company’s Global Safety, Intelligence, and Security (GSIS) team, which protects its people, facilities, and operations worldwide. A recent posting for a Protective Services Strategy & Operations Lead says the function covers intelligence, executive protection, investigations, travel and event security, security technology, security operations, environmental health and safety, and crisis management, and that it is growing quickly as Anthropic’s footprint, public profile, and threat surface expand.

The roles are senior and well paid:

Head of Protective Services: responsible for designing and running comprehensive executive protection programs and assessing risk for executive travel, events, and daily operations.
Executive Protection Program Manager: a since-closed San Francisco role listed at USD 275k that called for 10+ years of experience in executive protection.
Protective Intelligence Analyst: a now-filled role paying $175,000 - $220,000 per year, focused on all-source intelligence collection and analysis to identify, assess and prioritise physical security threats to Anthropic and its personnel.

Preferred backgrounds include advanced certifications in Executive Protection or Security Management and military or law enforcement experience.

Why Now: The AI Backlash Turns Physical

This spending reflects a broader industry shift. Threats targeting AI leaders and data centres jumped 7x between late February and May, according to the Wall Street Journal, citing threat intelligence firm Liferaft.

Anthropic has seen this directly. Days after an attempted firebombing at OpenAI CEO Sam Altman’s home, an intruder walked into Anthropic’s lobby to warn an executive was “going to be killed". San Francisco police responded to several threats against Anthropic staff and their counterparts at OpenAI.

The company says these precautions aren’t new. Anthropic has run round-the-clock security since 2024 and communicates regularly with employees about emerging threats, a company spokesman said.

Fear of job loss is a big driver of the anger. Palantir CEO Alex Karp has argued that fear of unemployment is feeding a backlash. Across the sector, several tech leaders are now moving with armed guards, while many have taken to silence on AI.

A Company Under Unusual Pressure

Anthropic’s profile has risen sharply this year. It is heading to public markets after investors valued it at $965 billion in May, when it raised $65 billion. It has also been in a public fight with Washington. In February, President Trump said he had directed all federal agencies to stop using the company’s technology after the company refused to drop limits on certain military uses.

Amodei’s own public warnings also keep him in the spotlight. Last month, he called for an industry-wide slowdown to allow time for safety research, industry standards and regulation to catch up.

What This Means for the AI Industry

The takeaway for founders and investors is that physical security is becoming a normal cost of running a frontier AI company. Analysts expect that if threats stay high, physical resilience and executive safety could become a standard part of AI infrastructure planning, rather than an exceptional response reserved for the most prominent companies.

For a company valued near a trillion dollars, $4 million is a small line item. But it shows how much the conversation around AI has changed: the people building the technology now need protection from the public they say they’re building it for.

InventiveLabs

Anthropic Spends Millions Protecting Dario Amodei as Threats Against AI Leaders Rise