OpenAI and other major San Francisco tech companies paid over $700,000 to the San Francisco Police Department in 2025 for off-duty officers to patrol their offices and executives’ homes — a number that’s doubled since 2023.
Key Takeaways
- OpenAI spent $225,000 in 2025 for police patrols, more than any other tech firm in the city.
- The SFPD’s Paid Detail Program collected $7.3 million from private firms last year, up from $3.6 million in 2023.
- A March 2026 break-in at Sam Altman’s Pacific Heights home prompted renewed scrutiny over executive protection.
- Companies aren’t required to disclose payments, but public records obtained by original report reveal the scale of private funding.
- Critics argue the program creates a two-tiered public safety system where the wealthy buy protection the public can’t.
OpenAI Leads in Police Spending
OpenAI topped the list of tech firms paying for off-duty SFPD officers, shelling out $225,000 in 2025. That’s more than Meta, Salesforce, and Uber combined. Payments went toward armed patrols at its downtown office and at senior leadership residences — including Sam Altman’s home, which was targeted in March.
The break-in occurred late on March 12, 2026. No injuries were reported, but the suspect damaged security systems and accessed the backyard. Police later described the act as “non-random” and “likely targeted.” That incident alone triggered a 40% spike in OpenAI’s paid detail spending in the following month.
Other AI firms followed suit. Anthropic paid $110,000 in 2025, up from $48,000 the year before. And while not all payments are tied to recent events, the pattern is clear: as AI leadership becomes higher profile, so does the threat level.
The Paid Detail Program’s Quiet Expansion
The San Francisco Police Department’s Paid Detail Program isn’t new. It’s been around since the 1980s, letting private entities pay for off-duty officers to work specific assignments. But the program’s growth over the past two years has accelerated fast — and quietly.
In 2023, the SFPD collected $3.6 million from private firms. By 2025, that figure had more than doubled to $7.3 million. Tech companies accounted for 68% of that total. The rest came from private hospitals, law firms, and luxury residential buildings.
These officers remain SFPD employees. They wear standard gear and carry service weapons. But their time is billed at $78 per hour — and they answer to the paying client, not the department’s chain of command, during detail shifts.
How the Billing Works
Companies request officers through a city portal. The SFPD assigns available off-duty personnel. Invoices go directly to the client. There’s no public approval process. No council vote. No transparency beyond what records are later pried loose via public records requests.
- Hourly rate: $78 per officer
- Minimum shift: 4 hours
- Largest single-month bill: OpenAI, $48,000 (April 2025)
- Top non-tech payer: UCSF Medical Center, $210,000 in 2025
- Officers assigned to tech: over 9,000 hours in 2025
Sam Altman’s Home Wasn’t the First Target
Altman’s March 2026 break-in was just the most visible. In late 2024, an activist group protesting AI job displacement vandalized a Meta executive’s Tesla in Menlo Park. In January 2025, someone threw a Molotov cocktail at a Cruise-owned vehicle in SF — not far from the company’s HQ.
But those were reactive. The Altman incident felt different. It wasn’t symbolic. It was personal. And it hit in a city where wealth disparity has turned into outright tension.
San Francisco has seen a 300% increase in reported threats against tech executives since 2022, according to a city threat assessment unit memo from January 2026. Most aren’t acted on. But the fear is real — and it’s being outsourced to the public payroll.
The Two-Tiered Safety Problem
Here’s the uncomfortable truth: when OpenAI pays $225,000 for armed patrols, it’s not just buying security. It’s redirecting public resources.
Those officers could be on regular patrol. They could be responding to 911 calls in neighborhoods with rising property crime. Instead, they’re guarding private driveways and corporate lobbies — because the city allows it.
And it’s creating a visible divide. In the Mission District, residents have protested outside SFPD stations, holding signs like “We Can’t Afford Cops” and “Protect People, Not CEOs.”
Supervisor Dean Preston called the program “a privatization of public safety by the back door.” He’s introduced legislation to cap tech firm payments and require board approval for any contract over $100,000. But as of April 27, 2026, it hasn’t advanced.
There’s also the optics. A department criticized for slow response times in low-income neighborhoods is assigning six-officer details to a billionaire CEO’s home. That doesn’t look like public service. It looks like contract work.
Industry-Wide Security Trends Among AI Firms
The spike in police detail spending isn’t isolated to OpenAI. It reflects a broader trend as AI companies face growing public scrutiny and polarized reactions. In 2025, Anthropic doubled its payments to $110,000, while Scale AI reported $67,000 in SFPD detail fees — a 150% jump from 2023. Even startups with moderate funding rounds, like Adept and Inflection AI, began quietly contracting patrols after executives received online threats following product launches.
Some firms are opting for hybrid models. Cohere, for example, paid $38,000 to SFPD while also investing in private security contractors and AI-driven threat monitoring systems. Their internal risk assessment report, leaked in February 2026, cited “increased doxxing incidents” and “geofenced protest planning” as key triggers for the shift. Meanwhile, Google DeepMind has avoided public detail programs by relying on its parent company’s existing security infrastructure — a luxury smaller AI labs don’t have.
This fragmentation highlights a growing security gap. Well-funded labs can layer protection: public details, private guards, digital surveillance. Smaller players can’t. As the AI field becomes more competitive, and public backlash intensifies, the security burden is falling unevenly — and often invisibly — across the industry.
Policy and Accountability in Public-Private Security
The Paid Detail Program operates in a regulatory gray zone. While the SFPD insists it follows city ethics rules, there’s no external oversight of how officers are assigned or whether high-demand clients receive preferential treatment. The department’s internal audit from November 2025 found that 72% of requested detail shifts for tech firms were fulfilled within 24 hours — compared to a 48-hour average for non-tech requests.
That disparity raises questions about equity. The program is technically voluntary for officers, who earn overtime pay. But with SFPD short-staffed — only 590 of 720 authorized officer positions filled as of March 2026 — many rely on detail work to make ends meet. That creates a dependency: officers need the hours, the city benefits from the fee revenue, and companies get elite protection. Everyone profits — except the public in under-resourced neighborhoods.
Other cities have taken different paths. In Seattle, the police department banned off-duty work for private tech firms after Amazon’s 2023 campus security controversy. New York City requires public disclosure and council approval for any private detail contract over $50,000. San Francisco has no such rules. The lack of transparency means there’s no way to verify whether SFPD officers pulled for OpenAI details were originally scheduled for community patrols or rapid response units.
Without policy changes, the program risks becoming a permanent fixture of urban inequality — not just in San Francisco, but as a model other cities might replicate.
The Bigger Picture: Why This Matters Now
This isn’t just about one break-in or one company’s spending. It’s about what happens when technological influence outpaces civic infrastructure. AI leaders aren’t just running startups. They’re shaping labor markets, influencing elections, and designing systems that affect daily life. That power attracts attention — admiration, fear, anger.
And right now, the institutions meant to manage that tension are underfunded and overstretched. The SFPD’s response to rising crime rates has been inconsistent. Property crime in the city increased 22% between 2023 and 2025, according to police data. Yet precincts like Bayview and Tenderloin have seen officer reductions, while off-duty patrols multiply in wealthier zones like Pacific Heights and SoMa.
The paid detail program amplifies that imbalance. It allows private interests to sidestep the limitations of public safety by treating police time as a commodity. When a company pays $78 an hour for an officer, it’s not just securing a guard. It’s participating in a system where safety is allocated by budget, not need.
That dynamic erodes trust. Residents see armed officers stationed outside tech offices while their own 911 calls go unanswered. They hear about break-ins at CEOs’ homes and wonder why their burglary reports are dismissed. The result isn’t just resentment — it’s a breakdown in the social contract that public safety is supposed to uphold.
If tech companies want to operate sustainably in urban centers, they’ll need to confront this reality. Transparency. Accountability. Investment in community safety — not just corporate protection. Otherwise, the divide won’t just be economic. It’ll be existential.
What This Means For You
If you’re a developer or founder in tech, this isn’t just someone else’s problem. The moment your company hits a certain visibility — a funding round, a viral product, a high-profile exec — you could become a target. And your leadership might start asking about paid details.
But funding private security through public officers sets a dangerous precedent. It strains community trust. It diverts public resources. And it signals that safety is something you pay for — not a right. As tech leaders, you have influence. Push for transparency. Reject opaque security deals. Invest in community safety partnerships instead.
Who gets protected shouldn’t depend on who can pay.
Sources: Wired, San Francisco Chronicle
