Data Privacy Vendor Launches Tool to Improve Security in AI Chatting
The rising concerns about AI security have driven a move by a data privacy vendor to release a redaction tool that can help firms reduce their customers’ and employees’ data privacy threats. Private AI‘s PrivateGPT platform automatically redacts over 50 types of personally identifiable information (PII) in real-time as users enter ChatGPT prompts. The platform interfaces with high-profile OpenAI chatbots and sits between a chatbot and a user, stripping health, financial, and personal data from the prompts before sending it through to ChatGPT. After the prompt, PrivateGPT re-populates the PII within the chatbot’s response to improve the user experience.
Privacy Risks Associated with ChatGPT
Every time a user types in data into the ChatGPT’s prompts, the AI model ingests the data and loads it into the service’s LLM data set, where it is used to train the next generation of the algorithm. However, data security experts and cybersecurity vendors fear that the data could be used later if proper data security isn’t installed for the service, thereby throwing caution to the wind. It is important to note that the AI model risks leaking sensitive corporate data that could be detrimental to an organization. Some of the concerns with AI models such as ChatGPT include presenting a black box of uncertainty, with regards to how and where a company’s data could end up stored, thereby upsetting the tight data privacy and security infrastructure that most companies rely on today.
The Importance of Secure Chatting
In a bid to address these lingering threats, Private AI has released an AI model that interfaces with OpenAI‘s ChatGPT that automatically redacts information that could be damaging to a client company, such as financial data, health data related to employees, social security numbers, transactions, and other identifiers. This applies to 100+ million ChatGPT users worldwide, primarily those in the software and tech ecosystems who may use AI to accelerate the delivery of code creation and analysis.
The recent security threat highlights the importance of secure chatting using AI models such as PrivateGPT. As such, developers worldwide seek to use AI models, but they need guidance from Engineering Management on the dos and don’ts of AI use to ensure data privacy is respected and maintained. The security threat is increasing as more employees and companies adopt AI models, and Private AI‘s PrivateGPT aims to close the gap and boost user confidence in using ChatGPT models.
Editorial
AI chatbots will remain relevant in organizations because they help to automate some functions. However, these AI models will be only useful if they are built with data privacy and security in mind. Companies that use AI chatbots should be mindful of the need for proper infrastructure setup to redact sensitive data. In conclusion, Private AI has introduced a timely solution in the form of PrivateGPT that can be used across OpenAI‘s ChatGPT chatbot to automatically redact 50+ types of personally identifiable information (PII) in real-time, thereby boosting the security of AI-enabled chatbot conversations.
Recommendation
Developers and companies that seek to adopt AI chatbots should understand the risks involved. Better still, regulators should provide clear guidance that managers can follow and a lead to efficient data privacy and security infrastructure setup. Companies should also review the IT policies, train employees and sign an AI policy. They should be vigilant of what questions the chatbot teaches their agents and what data they feed on. Finally, companies should review and assess the compliance risks regularly to ensure they remain on track with their data privacy and security objectives.
<< photo by Natã Romualdo >>
