After the turnaround and name change, the startup raises $140 million

The Israeli AI security company Alice, formerly ActiveFence, has completed a $140 million funding round led by Apax Digital. Globes has learned that the round was conducted at a valuation of approximately $800 million. The funding brings the total investment in Alice since its inception to $280 million, while at the same time, the company says it is approaching an annual recurring revenue (ARR) rate of $100 million.

The round also included participation from Samsung, the cyber company SentinelOne, MoreTech Ventures, and The Phoenix. Existing shareholders also invested, including CRV, Grove Ventures, Norwest Venture Partners, Highland, Vintage, ClalTech, and NFX. For comparison, in a funding round revealed in 2021, the company was valued at more than half a billion dollars.

The current round comes, as mentioned, about seven months after the company changed its name from ActiveFence to Alice, as part of a profound change in its operations. The company, originally founded to detect harmful content, fraud, and manipulation online, has in recent years made AI system security one of its main growth engines. Alice was included, then under its previous name, in the Globes list of promising startups for 2021.

From Google to Anthropic

Alongside the funding, Alice is expanding its client list. Among the prominent names: Google, Anthropic, Nvidia, Meta, TikTok, and Amazon. According to the company, its technology is used to protect services that reach more than three billion users, and it currently works with eight of the top ten AI labs developing advanced models. The fact that companies that compete with each other appear on the list is due to the type of product Alice offers. The company does not develop a general AI model of its own, but rather a testing and protection layer for the companies building the models and for organizations that integrate them into their products and systems.

Thus, before a new model is launched, Alice's researchers try to break its defense mechanisms and cause it to act against the rules set for it: to reveal information it is not allowed to disclose, to ignore instructions, or to perform an action it was not asked to perform. This process, known as "Red Teaming," allows for identifying weaknesses and adapting protections to the model, data, permissions, and tools to which it is connected.

After the system goes live, Alice continues to test it in the actual environment in which it operates. The organization can define which actions are allowed and prohibited, simulate attacks designed to expose data leaks or regulatory violations, and monitor in real-time the instructions sent to the system and the responses it returns. This activity is carried out through software products as well as through teams of researchers and engineers who work directly with clients. According to Alice, its AI security lab employs more than 150 researchers.

Also according to the company, the need for such a protection layer grows as AI agents gain broader access to information and organizational systems, and can not only answer questions but also perform actions independently. The 2026 International AI Safety Report stated that defense mechanisms are improving but still suffer from significant limitations: in some cases, they can be bypassed by rephrasing the request or splitting it into several stages.

Generative AI's market penetration led to a change in operations

Alice was founded in 2018 by Noam Schwartz, Yiftach Orr, Alon Porat, and Eyal Dykan. Under the name ActiveFence, it developed systems that helped social networks and internet platforms detect violent content, fraud, extremist propaganda, disinformation, and coordinated activity. This activity required the company to track how hostile actors change methods and attempt to bypass defense mechanisms.

With the launch of ChatGPT at the end of 2022 and the rapid penetration of generative AI into the market, it became clear to the company that the same knowledge could also be used to protect models and systems based on them. On one hand, criminal actors began using AI tools to generate fraud, phishing attacks, and misleading content on a larger scale. On the other hand, the companies developing the models needed a systematic way to test how they could be exploited. The knowledge accumulated in the original activity also forms the basis of Rabbit Hole, a database of attack patterns and harmful content collected from the real world. The database allows the company's researchers to test AI systems against methods already observed in the network and identify new versions of them. As the AI field took up a larger share of the company's activity, management decided that the name ActiveFence no longer reflected its business, and last January it completed the rebranding to Alice.

The funding will be used to expand Alice's development center in Israel, increase sales and marketing teams, and expand the team developing Rabbit Hole. Alice currently employs about 400 people, most of them in Israel, and the rest in offices in New York, London, and Hanoi.

"Our goal is for organizations to be able to trust their AI systems and know that they will operate as designed," said Noam Schwartz, CEO and co-founder of Alice. "To get there, you need to place experts in front of the models who challenge them every day."