OpenAI has warned employees to remain vigilant against potential espionage attempts, including so-called “honey traps”, in which foreign intelligence operatives allegedly use romantic or sexual advances to obtain confidential information, according to a Wall Street Journal report.
The reported security briefings outline a range of potential threats, from seemingly casual encounters with strangers to deliberate attempts to gain employees' trust through gifts, romantic overtures and repeated approaches.
One scenario described in the training involves a stranger befriending an OpenAI employee at a playground in San Francisco. The conversation initially centres on their children before shifting towards the employee's job, illustrating how seemingly harmless social interactions could be used to elicit sensitive information about the company.
Also Read | Layoffs Are Rising In Tech, But These Skills Are Helping People Unlock Better Pay
ALSO READ | AI Agents Emerge As Cyber Threat, Breaching Access Limits During Research, Warns CERT-In
Another example involves intelligence operatives repeatedly approaching a US diplomat in Russia through different individuals after their initial advances were rejected. Among those described were women carrying champagne and a young man.
The examples highlight how suspected intelligence operations may involve persistent approaches rather than a single encounter.
The reported protocols advise employees to keep a low profile outside the office, shield their devices from prying eyes on public transport, erase meeting-room whiteboards after discussions and remain alert to strangers who may be listening to conversations in public places, including cafes.
Employees at other major artificial intelligence companies, including Anthropic and Google, have reportedly received similar warnings.
Concerns over internal security have also drawn attention following a federal case involving Abdulrahman Alfozan, a Saudi national who worked at OpenAI.
US prosecutors have charged Alfozan with attempting to record audio using a device allegedly concealed behind a television in an executive conference room in San Francisco. He faces up to five years in prison and a $250,000 fine. His lawyer has described the allegations as “wholly out of character”.
ALSO READ | OpenAI Faces Fresh Legal Challenge As Eight US States Sue Over ChatGPT
The charge concerns an alleged violation of federal recording law. Alfozan has not been charged in this case with stealing trade secrets or acting as a foreign agent.
The security concerns reflect the strategic and commercial value of advanced AI technology, as companies seek to protect proprietary research, intellectual property and sensitive internal discussions.
With OpenAI and Anthropic reportedly linked to potential public listings at valuations exceeding $1 trillion, safeguarding their technology and confidential information has become an increasingly significant priority.
Essential Business Intelligence, Sharp Market Insights, Practical Personal Finance Advice, Daily Fuel, Gold and Silver Prices and Latest Stories — On NDTV Profit.
