For nearly a decade, one company kept the worst of the internet away from billions of people who never learned its name. That was the design. In January 2026, ActiveFence stopped hiding and stepped forward as Alice - and pointed its intelligence at the thing now embedded in everything: artificial intelligence.
A safety layer for communicative technology
Alice sells trust, safety, and security software to the platforms and AI labs that shape modern life. Its job is unglamorous and enormous: find the harm before it reaches a human being. That means scams and fraud, deepfakes and manipulation, child exploitation, violent extremism, coordinated disinformation - and, increasingly, the newer class of attacks aimed directly at AI models, such as prompt injection and jailbreaks.
The company's central claim is a philosophy as much as a product. Attackers do not think in code; they think in behavior. So Alice built its business around real-world abuse intelligence - a running, updated picture of how bad actors actually operate - and then taught software to recognize those patterns at machine speed.
Chief executive and co-founder Noam Schwartz frames it plainly: you cannot secure a model with code alone. The technical guardrails matter, but they are only useful when paired with an understanding of the human ingenuity trying to get around them. That combination - human threat intelligence plus automated detection - is what Alice packages and sells.
The rebrand from ActiveFence signals where the risk moved. The old name evoked a wall around user content. The new one, paired with product lines named WonderSuite, WonderBuild, WonderFence, and WonderCheck, points at the model itself as the thing now worth defending.
The platforms you use every day
Alice does not court the public - it courts the companies the public already relies on. Its named customers and partners read like a map of the modern internet: Amazon, TikTok and its parent Bytedance, Nvidia, the AI lab Cohere, and generative-image company Black Forest Labs. These are social networks, marketplaces, streaming services, gaming platforms, and the foundation-model labs training the systems everyone else builds on.
By Alice's own accounting, its technology helps protect around three billion people and underpins safety for seven of the ten largest AI foundation models. The scale is the point. When a safety layer works, the users it shields never notice it - which is precisely why so few people had heard the name ActiveFence before this year.
"Alice is the safety layer for all communicative tech. You cannot secure a model with code alone."
- Noam Schwartz, CEO & Co-Founder
Three fronts in one fight
Alice organizes its work around three broad threats, each of which has grown sharper as generative AI has spread.
Weaponized AI
Detecting scams, deepfakes, synthetic media, and manipulation created or amplified by AI - the abuse that now scales faster than any moderation team.
Interaction Security
Real-time guardrails and red-teaming against jailbreaks and prompt-injection attempts, stress-testing models the way an attacker would before they ship.
Model Integrity
Keeping model outputs safe, unbiased, and compliant, so that what a system produces stays inside the lines its makers intended.
Rabbit Hole and the Wonder line
At the core sits Rabbit Hole, which Alice describes as its adversarial intelligence engine. Built on a decade of real-world threat monitoring, it continuously analyzes billions of adversarial data points - the raw material of fraud rings, exploitation networks, extremist channels, and prompt-injection campaigns. The name is a knowing wink at how deep online threats can burrow.
Around that engine, Alice packages products for the people who actually ship software. WonderSuite is the umbrella; WonderBuild helps teams bake safety into AI applications, WonderFence supplies real-time protection and guardrails, and WonderCheck handles verification and integrity checks on outputs. For platforms with user-generated content, the company still offers the trust-and-safety intelligence that made its original name - harmful-content detection, influence-operation identification, and fraud prevention.
Where Alice guards you
Intelligence first, code second
The AI-security field has filled quickly with vendors offering scanners, firewalls, and evaluation tools - names like HiddenLayer, Robust Intelligence, Lakera, and Protect AI, alongside the in-house safety teams at every large platform. Most approach the problem as an engineering task: instrument the model, catch the anomaly, patch the gap.
Alice's wager is that engineering alone arrives too late. Its edge is the accumulated, human-curated understanding of how malicious behavior emerges and spreads - eight years of watching abuse mutate across social platforms before it ever reached generative AI. That intelligence is difficult to buy and slow to build, which is the moat the company is counting on.
The 2023 acquisitions of Rewire and Spectrum Labs, two trust-and-safety startups, deepened that library of behavioral knowledge. And because Alice already sat inside the largest platforms, it had a front-row seat as those same customers began asking for help securing the AI models they were racing to deploy. The pivot to model security was less a leap than a following of demand.
Selling trust by subscription
Alice runs a business-to-business SaaS model: platforms, foundation-model labs, and enterprises building AI features pay for its intelligence, detection, and security software, typically through annual contracts and API-based integrations. The expertise is concentrated in a team of former intelligence, defense, and trust-and-safety practitioners split across offices in New York and Tel Aviv - people whose careers were spent studying how systems get abused.
Who built it
Alice was founded in 2018 by Noam Schwartz, who serves as chief executive, together with co-founders Alon Porat, Eyal Dykan, and Iftach Orr. Schwartz has become the public face of the rebrand, arguing that effective AI security demands a deep understanding of human-driven abuse rather than technical controls alone. The founding team grew the company from roughly 60 employees to around 200 in the single year surrounding its 2021 Series B.
From a wall to a layer
ActiveFence founded
Noam Schwartz and co-founders launch the company to detect harmful content and threats across online platforms.
$100M Series B at a $500M valuation
CRV and Highland lead the round, revealing an earlier Series A and fueling rapid team growth.
Acquires Rewire and Spectrum Labs
Two trust-and-safety startups join, deepening the company's content and moderation capabilities.
Rebrands as Alice
ActiveFence becomes Alice, repositioning as the safety and security layer for AI and social platforms and unveiling the Rabbit Hole engine.
The infrastructure nobody sees
Alice occupies a specific rung of the AI stack: not the model, not the application, but the safety layer between them and the outside world. As regulation tightens and public trust becomes a competitive asset, that position looks less like a cost center and more like table stakes. The internet grew up without such a layer, and two decades of fraud, abuse, and disinformation were the bill. Alice's bet is that AI does not have to repeat the lesson - and that this time, safety ships alongside the product rather than years behind it.
What it has to show for it
Eight years of quiet work leaves a record. Alice's technology now helps protect roughly three billion people and secures seven of the ten largest AI foundation models - a footprint most consumer brands never reach. It has raised about $140 million in total, crossing a $500 million valuation at its 2021 Series B, and folded two specialist startups, Rewire and Spectrum Labs, into its intelligence library. The 2026 rebrand and the launch of the Rabbit Hole engine mark the moment the company decided its next decade would be defined by AI, not just the open internet.
Quick answers
Is Alice the same company as ActiveFence?
Yes. Alice is the new name for ActiveFence, which rebranded in January 2026 while expanding from online content safety into AI model security.
What does Alice do?
It provides trust, safety, and security software that detects and blocks online harms - fraud, deepfakes, child exploitation, disinformation, prompt injection, and jailbreaks - for both platforms and AI systems.
Who uses Alice?
Major platforms and AI labs including Amazon, TikTok, Nvidia, and Cohere. Its technology reportedly helps protect around 3 billion people and secures 7 of the 10 largest AI foundation models.
Who founded Alice and when?
It was founded in 2018 by Noam Schwartz (CEO) along with co-founders Alon Porat, Eyal Dykan, and Iftach Orr.
How much funding has it raised?
About $140M to date, including a $100M Series B in 2021 that valued the company at roughly $500M.
Links, news & profiles
Figures such as users protected, models secured, funding, valuation, and revenue are drawn from public reporting and company statements; some are self-reported or third-party estimates and are approximate.