IBM Automation’s Chief Technology Officer, Jerry Cuomo, recently penned a blog post that delves into the potential risks tied to deploying ChatGPT for enterprise purposes.
Cuomo highlights crucial areas of concern that businesses ought to ponder before integrating ChatGPT into their operations.
Despite these concerns, Cuomo ultimately asserts that ChatGPT is suitable solely for non-sensitive data, emphasizing that once information is fed into ChatGPT, its usage becomes opaque and uncontrollable.
Cuomo underscores the inadvertent likelihood of data leakage, posing a legal predicament for companies if confidential customer, partner, or client data seeps into ChatGPT’s training dataset and subsequently becomes public.
Intellectual property concerns also loom large, and Cuomo warns against the inadvertent violation of open-source agreements due to such data leakage.
In line with the IBM blog post, Cuomo states that if proprietary third-party or internal organizational data is inputted into ChatGPT, it morphs into a part of the chatbot’s data framework and could potentially be shared with others seeking relevant insights.
Following the article’s publication, Cointelegraph sought input from OpenAI on the matter.
READ MORE: Silvergate Bank Undergoes Executive Shake-Up Amid Crypto Transition and Legal Challenges
A public relations representative responded via email, asserting that the data would not be shared with other users posing relevant queries.
The representative further directed attention to the existing documentation elucidating ChatGPT’s privacy safeguards, including an explanation of how web users can deactivate their chat history.
OpenAI’s ChatGPT API, as disclosed, has data sharing deactivated by default.
However, critics have voiced concerns that conversations on the web-based iteration are automatically archived.
Users must actively opt out of saving their conversations, despite the utility of the feature in picking up where discussions left off, and the utilization of their data for model training.
Presently, no alternative exists for conserving conversations without consenting to data sharing.
In summary, IBM Automation’s CTO, Jerry Cuomo, has highlighted numerous vulnerabilities inherent in deploying ChatGPT for enterprise applications.
These vulnerabilities encompass inadvertent data leakage, potential legal repercussions, and intellectual property risks.
OpenAI’s response underscores the commitment to data protection, but critics have pointed out discrepancies between the web version’s conversation saving feature and the ability to opt out of data sharing for training.
Other Stories:
Former US President Donald Trump’s Ethereum Wallet Surges to $2.8 Million
Stellar Development Foundation Invests in MoneyGram International
Coinbase’s ‘Stand with Crypto Alliance’ Gains Momentum Amidst Lawmaker Engagement Push