Four months after emerging from stealth, Tel Aviv-based Aim Security has raised $18 million in a Series A Round led by Canaan Partners with participation from YL Ventures. Additional backers include the founders of WIZ and leaders from Palo Alto Networks, Proofpoint and Google.
Artificial Intelligence (AI, and especially generative AI, gen-AI) threatens to be the biggest single influence on business since the emergence of cloud computing – but the effect of gen-AI on the corporate risk surface is still largely unknown and widely misunderstood.
For example, the security of in-house developed chatbots and virtual assistants, and in-house and third-party productivity tools developed on the back of gen-AI using known and unknown data lakes is a growing concern. Gen-AI is expanding the attack and threat surfaces in ways that are not necessarily understood by security teams. This concern can delay the implementation of gen-AI productivity and harm business competitiveness or expose rapid adopters to both bad actor and compliance risk.
A primary example of ‘new’ threat introduced by gen-AI applications is prompt injection leading to jailbreaks – that is, the generation of an unintended response from the AI application through specially crafted enquiries or prompts. The purpose is to bypass the gen-AI guardrails – that is, create a jailbreak.
The guardrails are effectively filters on the acceptable use of the application. Studies have shown that in-house developed filters are unlikely to stop user or bad actor misuse (see Your Customer Chatbot is Almost Certainly Insecure).
While the possible effect of malicious prompt injection will vary from product to product, it could range from mischievous (causing hate speech or simple misinformation leading to brand damage); the injection of new data to skew future outputs (poisoning); abuse of privacy and contravention of privacy and other vertical-specific regulations; and even access to functionalities not intended for user access. As the complexity of AI products increases, so too will the risks they bring.
Aim Security intends to provide the expert support necessary for companies to implement AI products with confidence. It was founded by Matan Getz (CEO) and Adir Gruss (CTO) in 2023. Both are alumni of the IDF intelligence Unit 8200 where they broke and built AI systems – including developing a major data science defensive AI project for the unit.
With Aim Security, “CISOs now understand that they can comfortably adopt AI across their entire organization without jeopardizing their company’s assets,” said Getz. The new funding will have the dual purpose of bolstering Aim’s product development and fueling its go-to-market efforts across the globe.
Drew Robertson, CISO at existing customer Finance of America, explained the problem and Aim’s solution. “The stringent regulations of the finance industry and concern for our customers’ data make gen-AI adoption challenging. Aim secures the entire breadth of our gen-AI use, no matter where it is applied – public SaaS apps, enterprise chats, or our own internal developments – and helps us place and enforce the guardrails that fit our industry regulations and unique needs.”
Aim Security emerged from stealth in January 2024 with $10 million seed funding. This Series A funding brings the total raised to date to $28 million.
Learn More at SecurityWeek’s AI Risk Summit
Related: Simbian Emerges From Stealth With $10 Million to Build Autonomous AI-Based Security Platform
Related: Shadow AI – Should I be Worried?
Related: Cyber Insights 2024: Artificial Intelligence
Related: Israeli Startup Gets $5M Seed Capital to Tackle AI Security
