Prompt Security Bags $18M to Protect Against GenAI Vulnerabilities
With great GenAI comes great (security) responsibility
GenAI, when deployed responsibly, makes life easier on enterprises.
Conversely, a loosey-goosey approach to GenAI deployment can lead to financial, reputational, and even legal harm.
Enter Prompt Security, a trailblazing solution in the burgeoning GenAI Security space and (pats back) the newest addition to our Ridge Familia. Today, Prompt raised its $18M Series A to protect enterprise companies from the myriad security threats posed by GenAI applications.
Here is the lowdown on Prompt, and how they’re enabling enterprises to use GenAI applications worry-free.
A new security frontier
The fervor around GenAI has reached color TV and first-generation iPhone proportions. Along with this excitement and endless potential, however, comes a laundry list of security concerns for enterprises.
GenAI security impacts companies both internally and externally. Employees’ misuse of GenAI applications can lead to data breaches. Enterprises deploying GenAI products to customers carry an even larger burden, including the possibility of lawsuits and Hindenburg-esque reputational damage.
Internally, employees may believe they’re harmlessly using ChatGPT to write marketing copy, or DALL-E to generate images for a Powerpoint deck. In actuality, they could be inadvertently sharing proprietary data with public LLM providers. Some companies try to mitigate this data privacy risk through firewalls and browser protections, but they’d have better luck Scotch taping a water leak. Such an approach can hamstring productivity and deem GenAI tools less efficient. Regex-based solutions aren’t any better, as they lack sufficient context to effectively redact sensitive data.
Externally, companies deploying GenAI solutions face compliance and data protection issues, but also nascent security risks like prompt injection, prompt leakage, and jailbreaks. The lack of a tried-and-true playbook against these threats has CISOs and CIOs waking up in cold sweats. Customer churn is one thing, but many enterprises are legally bound by Service-Level Agreements (SLAs) that, if violated, could lead to financial penalties and legal action. Bereft of real-time monitoring, threat intelligence, and rapid response capabilities, meeting the requisite security and compliance standards for GenAI products is a Sisyphean task.
The sultan of GenAI security
Daunting as this new GenAI security landscape may seem, Prompt Security provides everything companies need to protect themselves (and their users). The best-in-class designation gets thrown around a lot these days but, after surveying the GenAI security space for quite some time, we can definitively say Prompt is top dog.
Here are just some of the GenAI security threats Prompt neutralizes:
- Shadow AI: Shadow AI, or when employees use an AI application without the permission of their security department, can expose sensitive data and cause compliance issues. Prompt Security, which estimates that the average organization uses at least 50 GenAI tools weekly, detects Shadow AI dynamically in real time.
- Jailbreaking: Jailbreaking is when prompts are engineered to exploit model biases and generate outputs that contradict the intended behavior or original purpose of an LLM. Prompt Security’s non-stop monitoring of each prompt and response shields companies from reputational damage and poor customer experiences.
- Toxic or harmful content: A jailbroken LLM can expose users to inappropriate, toxic or off-brand content, potentially leading to reputational or legal damage. Prompt Security’s extensive moderation filters scrutinize every response generated by the LLM powering an application before it reaches a customer or employee.
- Prompt leak: A form of prompt injection in which an LLM mistakenly reveals its internal logic. This can lead to the disclosure of intellectual property, as well as more harmful prompt injections downstream. Prompt Security monitors each prompt and response so GenAI apps don’t reveal their assigned instructions, policies, or system prompts.
Implementing Prompt Security is no sweat. In the cloud or on-prem, Prompt easily slides into any security stack — and is deployed within minutes — so companies can fortify their GenAI security defenses in a jiffy.
A GenAI dream team
It all starts at the top, and with co-founders Itamar Golan (CEO) and Lior Drihem (CTO) manning the controls, Prompt Security boasts a winning combination of domain expertise, leadership chops, and product innovation.
Itamar, who studied both computer science and neuroscience at Tel Aviv University, previously led the Data Science and Machine Learning Engineering teams at Orca Security. He knows the ins and outs of data infrastructure, and architecting top-notch products comes as easy as tying his shoes. Itamar has also managed large data science teams in the past, experience that will pay dividends when Prompt Security scales into a GenAI security behemoth.
Lior, Prompt Security’s technical wizard, led Innovation at Orca Security, where he crossed paths with Itamar. Prior to Orca, Lior spent 18 years at Check Point Software Technologies, working his way up through the project management and software development ranks before serving as Director of Innovation. Lior, who for decades has remained an influential product leader in the security space, worked on VPN products at Check Point and incubated the company’s first browser-based security solution.
Ensuring they stay on top of the latest GenAI security trends, Itamar and Lior are also both active members of the Open Worldwide Application Security Project (OWASP) — essentially the Justice League of software security. Specifically, they help compile the OWASP Top 10 for Large Language Model Applications, a document outlining the most dangerous security threats for those deploying and managing LLMs.
Securing the future of GenAI
Security leaders no longer need to choose between using GenAI and protecting data and compliance. Prompt Security gives them the best of both worlds.
Ridge is proud to support such a timely and pioneering solution. In doing so, we team up with our friends at Jump Capital and Hetz Ventures, as well as Okta and even one of our own portfolio companies: F5!
A hearty congrats to Itamar, Lior, and everyone at Prompt Security on the big news. Here’s to the next milestone. 🚀
Are you a GenAI app developer? Give Prompt Fuzzer a spin. The free open-source tool is a surefire way to test the security of your prompts!
