OpenAI and Anthropic employees are instructed to hide their identities in public, to guard against tricks to extract information such as eavesdropping or emotional exploitation.

According to New York Post, OpenAI is asking employees to clean whiteboards in meeting rooms after use, not to wear anything with the company logo when out of the office, and to use anti-peeping screen protectors when traveling by public transportation. The company also holds training sessions to help employees be wary of spies who approach money, friendship or romance.

Anthropic, OpenAI's big rival in the race to develop artificial intelligence, also bans almost all company-branded items, reminding people to hide their employee ID cards when leaving the office. An Anthropic spokesperson told WSJ the company has maintained 24/7 security operations since 2024 and regularly communicates with employees about emerging threats.

"I strongly support this strategy," said Miao Hong, partner at Silicon Valley Future Capital, an investor in OpenAI and Anthropic. He said that he has seen people working in the AI ​​field constantly being approached by people who want to extract internal information.

"Everyone wants to learn about developments and information at the forefront of technology. They want first-hand information, they want to talk to employees, but employees generally avoid discussing anything," he commented.

Phil Venables, a partner at venture capital firm Ballistic Ventures and former chief information security officer at Google Cloud, said corporate espionage has existed since the emergence of intellectual property that can be stolen. The concentration of so much value in a small number of AI companies, such as OpenAI and Anthropic, makes them attractive targets.

According to WSJ, during security training sessions, OpenAI warns that opponents or spies from other countries use many tactics to extract information, from eavesdropping in coffee shops to bribery, even emotional exploitation. The company cited the case of a stranger trying to befriend an employee at a playground in San Francisco. This employee brought her child along, and strangers often came with the same child. However, this person is actually a spy and intentionally added questions about work while chatting.

OpenAI also told another example to illustrate the "emotional trap", implying a similar situation could happen to them. Accordingly, to seduce an American diplomat, a foreign intelligence organization sent a blonde woman to bring wine to his hotel room. When refused, the organization continued to send a red-haired woman, then a young man.

Meanwhile, the father of an Anthropic employee said that his son and his colleagues limit talking about work after work, usually only talking about sports, hobbies, climbing activities or eating. Sharing with New York Post, he said that his job especially makes it difficult for his son to find someone trustworthy. "Whether going to a party, finding your 'other half' or making friends, people focus on your company and assets instead of your qualities and values," he explains.

For AI business leaders, things are even more complicated. According to JPT Security company, espionage activities targeting high-ranking executives often begin with techniques that manipulate psychology and social behavior. Spies can create reasons to approach leaders, employees, and even their family members.

Business leaders also face a high risk of being attacked. The home of Sam Altman, CEO of OpenAI, in San Francisco was the target of two attacks in April. In the first incident, the suspect threw a gasoline bomb into his house and then threatened to burn down OpenAI headquarters. Two days later, two other suspects opened fire outside the house. All three have been arrested and are awaiting trial.

Meanwhile, the IPO documents filed by Anthropic with the US Securities and Exchange Commission (SEC) revealed that the company spent $2.7 million on "personal security services" of CEO Dario Amodei.

According to analysis by data company Equilar, in 2025, 38.1% of technology companies in the S&P 500 group have announced spending on senior executive protection, a significant increase compared to 26.8% in 2021.

Thu Thao summary