Less than three weeks after Samsung lifted its ban on employees using ChatGPT, the chaebol has reportedly revealed its own secrets — including sensitive semiconductor information under development — on at least three occasions.
The ban was originally intended to protect company data, but was lifted on March 11 in a bid to boost productivity and keep employees busy with the world’s latest cool tech tools.
According to a Korean local media report, those secrets now leaked include device measurements and earnings data from the conglomerate’s device solutions and semiconductor business unit.
An employee told the local outlet that they copied the entire problematic source code of a semiconductor database downloader, entered it into ChatGPT and asked for a solution.
Another uploaded code used to identify broken devices, and a third uploaded recordings of a meeting to automatically create minutes.
ChatGPT’s FAQ states, “Your conversations may be reviewed by our AI trainers to improve our systems.”
Samsung’s secrets could therefore be accessible to OpenAI – an entity with more than a passing interest in the Korean giant’s technology and affairs.
After the incidents were discovered, Samsung reportedly took “emergency measures,” including limiting upload capacity to 1024 bytes per question.
“If a similar accident happens, even after emergency measures are taken to protect information, access to ChatGPT on the corporate network may be blocked,” Samsung bosses reportedly warned employees.
The chaebol had already warned its employees not to disclose proprietary information when it lifted ChatGPT’s blanket ban. OpenAI also warns against uploading sensitive data.
As of March 1st of this year, OpenAI guidelines dictate that API users must sign in to share data to train or improve their models, while non-API services – like ChatGPT or DALL-E – are managed by a user require them to opt-out to avoid having their data used.
“We remove all personally identifiable information from data that we plan to use to improve model performance. We also only use a small sample of data per customer in our efforts to improve model performance,” claims OpenAI.
According to local media reports, Samsung is now considering building its own internal AI service to prevent further incidents.
the reg has asked Samsung to confirm the details of this story but had not received a response at the time of writing.
Accordingly The Korea TimesThe incident has rocked domestic tech companies, including chipmaker SK hynix and consumer hardware company LG Display, which are now working on guidelines for using AI chatbots.
SK Hynix has reportedly blocked the use of chatbots on its internal network, and employees who wish to use the service must obtain security clearances. LG Display has apparently decided to launch an awareness campaign so that its employees understand how to protect company secrets.
Step one is probably “Don’t upload it to someone else’s website”. ®