Preamble, a Pittsburgh based startup, launched its artificial intelligence security software as a service this week. The service places "guardrails" on artificial intelligence chatbots in the workplace.
“This has been a long time coming building up this platform,” CEO Jeremy McHugh said. “Knowing that AI is such a complex solution it took a lot to understand just what the customer needs, how to simplify it for them and the easiest way to start using it.”
The use of artificial intelligence in the workplace has increased significantly as access to technologies like ChatGPT has become more readily available. According to OpenAI, ChatGPT’s developers, the software currently has over 100 million weekly users and generated 1.63 billion website visits in February of 2024.
According to the Information Systems Audit and Control Association, only 10% of businesses have a “formal comprehensive policy in place” on AI, despite 40% of businesses knowing of their employees using the technology and 35% unsure of the use in the workplace.
Preamble integrates chatbots through its software to integrate workplace safety, security, privacy and general limitations on the technology.
“The AI providers trying to do this for everybody doesn’t help really anybody,” McHugh said. “By allowing the companies to dictate these types of controls because they all have their own internal ethics and mission statements and values that they want to implement and reflect in their own policies, that’s where we give them the control.”
Privacy is a large concern for companies working with AI. OpenAI has publicly acknowledged data breaches on more than one occasion. One of the features of Preamble’s service is to limit the submission of proprietary and personal identifiable information.
“If privacy is a concern of a company of its employees accidentally submitting some content that has sensitive data they can have those rules that blocks them from being sent to a third-party AI service,” McHugh said.
The company, which operates out of a coworking space downtown, currently offers full access to OpenAI and Mistral AI and plans to integrate more services.
“On a regular basis, we will keep adding different AI models that are being requested and being used more,” McHugh said. “For some enterprises they may want to have their own custom models and we can provide support for those as well.”
Public perception of AI varies, in part due to the complexity of the technology. The Pew Research Center found that 52% of Americans are “more concerned than excited about AI in daily life” but that only 30% of Americans could recognize all of the examples of current implementations in daily life.
“We’re headed towards an AI-centered world,” McHugh said. “I think it’s going to be implemented in such a way that there are AI-centric services and applications where you might not even know that it is AI.”
Beyond guardrails that his company can implement, McHugh said that there needs to be a larger discussion on automation and the widespread use of AI, as well as the impact that a cyberattack could have on automated services.
“If everything is automated and there are no people there then what’s the backup plan if these processes are failing?” McHugh said.
Governing bodies have differing views on AI and how to regulate it, as well as what constitutes appropriate use. Pennsylvania has taken a more open stance on AI and is the first state to partner with OpenAI for a pilot program. The pilot program gives government employees access to a version of ChatGPT Enterprise, a business oriented version of the chatbot. McHugh said that he “would love to work with” the government to help put guardrails on the technology.
Preamble is currently seeking funding, but McHugh said that “the focus is on the product” and “solving customer problems.”
