The Alice, a company specialized in security and protection of artificial intelligence systems, announced this Tuesday (25) a round of US$ 140 million led by Apax Digital funds. The investment raises to US$ 280 million the total raised by the company and will be used to expand testing, monitoring and defense technologies for AI models.
The round also had participation from Samsung, SentinelOne, Maj Invest, MoreTech and Phoenix Insurance, in addition to investors already present in the business, such as CRV, Highland Europe, Norwest Venture Partners, NFX and Grove Ventures. Apax Digital will join the company's board.
Alice says it currently works with eight of the ten leading AI model labs and is close to reaching US$ 100 million in annual recurring revenue (ARR). According to the company, revenue from its AI-focused operation grew more than 500% in the last two years.
Security for models before and after launch
Alice's platform operates at different stages of the development and operation of AI systems. Before the launch of the models, company researchers simulate malicious prompts and tasks executed by autonomous agents to identify unexpected behaviors and vulnerabilities.
Among the problems evaluated are attacks of jailbreak, used to bypass model restrictions, and prompt injection, a technique that attempts to manipulate AI systems through instructions inserted into processed data.
The company cites work with labs such as Anthropic, Google and Cohere. After model deployment, the technology also allows companies to establish their own policies, simulate attacks and monitor inputs and responses to detect risks of data leakage, rule violations and unwanted behaviors.
Alice maintains a laboratory with more than 150 researchers dedicated to studying forms of manipulation, failures and attacks against artificial intelligence systems. The company also states that its technology is used to protect more than 3 billion people on digital platforms.
Investment will be directed to the platform and to Rabbit Hole
Part of the new funding will be applied to expanding Rabbit Hole, a proprietary database that gathers information about adversarial and harmful content observed in digital environments. The company uses this history to identify patterns that may reappear in attacks against AI systems.
Alice also plans to invest in tools that test and monitor models throughout their lifecycle and to expand its commercial structure focused on foundational model labs and companies that are incorporating artificial intelligence into their operations.
Founded in 2018 and previously known as ActiveFence, Alice is headquartered in New York and Tel Aviv. The company began directing the experience accumulated in combating fraud, extremism and coordinated manipulation on digital platforms toward the security of AI models and agents.
