If someone is suddenly very interested in you and your job at OpenAI, you may want to ask yourself why, and be very wary. That's the warning the OpenAI management is passing on to its employees.
In a Hollywood-esque warning, OpenAI has been telling its employees that foreign spies could try to seduce them to get at company secrets, according to a WSJ report.
The tactic is known as a "honey pot", where an agent uses sex or romance to get close to a target.
The Playground Spy
Not all the warnings are sexy or steamy though.
One scenario OpenAI reportedly shares is about a stranger befriending an employee at a San Francisco playground.
They chat about the kids, then slowly the conversation steers to what the employee does for work. The stranger, it turns out, is a spy.
The Champagne Test
To drive the point home, trainers reportedly tell a story about a US diplomat in Russia.
Spies first sent a blonde woman to his hotel room with champagne. He turned her away. So they sent a redheaded woman. And then a young man.
The message and learning for the staff is: they are relentless and they will keep trying.
The Spy-Proofing Handbook
Employees have been reportedly told to follow certain standard operating procedures (SOPs) such as:
-- Skip company-logo clothing in public
-- Use privacy screens on phones and laptops on public transport
-- Wipe whiteboards clean after meetings, in case cleaning or kitchen staff are compromised
-- Watch for people eavesdropping in cafes
Anthropic and Google employees have reportedly received similar warnings.
And Then, This Week, It Got Real
Abdulrahman Alfozan, a Saudi national who worked at OpenAI, was charged with trying to record audio using a device planted in an executive conference room in San Francisco. The Justice Department says the device was plugged into the back of a TV and hidden in toilet paper. He faces up to five years in prison and a $250,000 fine.
His lawyer says the allegations are "wholly out of character." Alfozan is charged only with violating a federal recording law, not with stealing trade secrets or acting as a foreign agent.
FBI Director Kash Patel said protecting AI secrets is "one of our highest counterintelligence priorities."
Why So Much Fuss?
Because the stakes are enormous.
OpenAI and Anthropic are each expected to seek valuations above $1 trillion in their IPOs (initial public offerings), and AI is seen as a technology that could reshape economies and even warfare.
Former OpenAI and Anthropic researcher, Jacob Coxon, who became famous for whistleblowing on the AI labs' laggard attitude towards AI safety is on record saying that he's 90% certain that spies have already infiltrated the big AI labs.