A bugged OpenAI conference room is the latest in a series of espionage suspicions and trade-secret lawsuits at tech companies.
Fears of corporate espionage are sweeping Silicon Valley.
The FBI last week arrested an OpenAI engineer who is suspected of spying, saying the employee, a Saudi national, hid a recording device in a conference room at the company’s San Francisco headquarters. This month, OpenAI fired three safety researchers, saying it could not trust them (opens in new tab) due to alleged misconduct, including reportedly sharing confidential information (opens in new tab) with external safety evaluator METR — details of which the researchers dispute (opens in new tab) . And in July, Apple accused (opens in new tab) OpenAI of stealing trade secrets.
As AI companies compete to be the main global supplier of powerful technology they say will transform economies, fears of leaks, theft of trade secrets, poaching, and espionage are rising. Top players leapfrog one another regularly and have few technical advantages at a given time. Silicon Valley’s long-standing culture of highly paid workers moving between competitors has put pressure on labs trying to keep any edge they can.
“This is the second gold rush for California,” said Dylan Wiseman, a lawyer at Buchalter who specializes in trade secrets. “There is going to be friction among these players in the field for a long time.”
Tech companies typically sue employees they suspect have taken information. In recent years, Tesla (opens in new tab) , Meta (opens in new tab) , and Scale AI (opens in new tab) have done exactly that. Last year, xAI accused (opens in new tab) a former engineer of taking confidential information its chatbot Grok after accepting a job at OpenAI. A judge dismissed (opens in new tab) xAI’s lawsuit against OpenAI, alleging that it poached employees to get information. OpenAI leaders have warned (opens in new tab) employees spies trying to befriend them at playgrounds, eavesdrop on conversations at coffee shops, and even seduce them — echoing similar warnings at Anthropic and Google, the Wall Street Journal reported. Federal prosecutors have also charged (opens in new tab) several former Google employees with stealing trade secrets.
OpenAI engineer booked one-way Saudi flight before arrest
OpenAI engineer booked one-way Saudi flight before arrest
Ex-Mayor Art Agnos takes stand against Big Tech in District 10 race
Ex-Mayor Art Agnos takes stand against Big Tech in District 10 race
Anthropic uses AI to dodge customer complaints. A new law may force a change
Anthropic uses AI to dodge customer complaints. A new law may force a change
Allegations of corporate espionage aren’t limited to the biggest AI labs. HR tech platforms Rippling and Deel are embroiled in a high-drama spy lawsuit that began last year over an alleged mole Deel had recruited within its competitor. AI insurance startup Corgi filed lawsuits (opens in new tab) against competitors and former employees, alleging stolen technology and attempts to gain other proprietary information.
The reason for the increase? A mix of timing and state law, according to Wiseman, who said he is working on five trade-secrets lawsuits involving tech companies. Many of the startups that emerged from the AI boom within the last few years have grown up enough to have significant funding and viable competitors — and their early employees are beginning to leave. California law doesn’t allow most noncompete clauses, so workers are generally free to leave their companies and immediately begin working for a competitor. That leaves trade-secrets law — which protects business information that holds economic value and competitors can’t easily attain — as one of the few means of recourse to protect internal information.
“There’s going to be a significant amount of continuing litigation involving AI firms,” Wiseman said. “It’s like drinking out of a firehose, being a lawyer.”
As the global race for AI supremacy intensifies, companies are looking for spies not only in Silicon Valley but across oceans as foreign workers attempt to get U.S. jobs through the use of deepfake technology.
The technology, which uses AI to generate fake images and voices of real people, has been around for years, but it was expensive to use. Starting last year, the improvement of open-source models that run locally on a computer, rather than in a data center, made producing deepfakes much cheaper and more accessible. It has been a key tool used by bad actors to conceal their identities in video calls, including job interviews, to infiltrate companies and steal information.
“That singlehandedly opened the floodgates,” said Ben Colman, CEO of deepfake detection company Reality Defender.
Most of the several dozen Fortune 500 companies using Reality Defender’s technology believe they have been a target of corporate espionage from adversarial countries, he said. It’s difficult to determine the intent of job seekers. Some foreign workers, especially from North Korea, attempt to work remote American jobs for high salaries that contribute to regime funds — though those workers have been tied to data theft (opens in new tab) as well.
In hiring for a top tech role, one financial services firm using Reality Defender found indications of espionage in three of its four finalists, Colman said, declining to name the company. Two were made-up personas that still passed background checks, and one was a real person who, when offered the job directly, said they had never applied, Colman said.
Pindrop Security, which also detects deepfakes, said a person who was likely a North Korean agent applied for a remote senior software engineering position in the U.S. last year. The individual “appeared well prepared and answered questions confidently in a virtual interview,” Pindrop CEO Vijay Balasubramaniyan said, but the company later linked their IP address to North Korea. Across its clients, Pindrop has identified signs of North Korean ties in 1 of every 47 applicants it has reviewed this year, a sevenfold increase from the year before.
“HR has not traditionally had as much tech innovation as everyone else in the company,” Colman said. “For the first time, HR is in all of our cyber conversations.”
More the author
Rachyl Jones Tech Reporter · Rachyl [email protected] @RachylJones (opens in new tab) @rachyljones (opens in new tab) rachyl.22 Send a tip Rachyl Jones is a tech reporter at The Standard, focusing on AI.
@RachylJones (opens in new tab)
@rachyljones (opens in new tab)
Rachyl Jones is a tech reporter at The Standard, focusing on AI.
The full story
This article is one source in a clustered incident — the cluster page carries the summary, timeline and every other outlet covering it.
