Alice raises $140 million to break AI models before attackers do
Two big rounds landed tonight — one for the people who attack AI models for a living, and one for the trucks that drive themselves to the store.
Alice, the AI trust-and-safety company until this year known as ActiveFence, has raised $140 million led by Apax Digital Funds — and the numbers suggest guarding models has become a real business rather than a compliance line item. Most frontier labs already pay for its adversarial testing ahead of release, according to the company, and annual recurring revenue is approaching $100 million after the AI side of the business grew more than 500% over two years. Calcalist reported a valuation between $700 million and $800 million, though Alice hasn't confirmed it. Samsung and SentinelOne joined the round, bringing total funding to $280 million.
What buyers actually get: Alice spent its first decade logging fraud rings, extremist networks and coordinated manipulation campaigns into a dataset it calls Rabbit Hole — arguably the largest collection of real-world abuse patterns in existence. Those stored attacks now double as pre-release test cases fired at models from Anthropic, Google and Cohere, while enterprise customers set their own usage policies, simulate attacks against them and watch production traffic for gaps. CEO Noam Schwartz frames the problem plainly: the industry is "democratizing capabilities before we've democratized the defenses."
Our take: the telling detail isn't the round size, it's the customer list. Labs that employ their own red teams are still paying an outside firm to break their models — the same reason companies with strong security teams still buy penetration tests, because in-house eyes go soft on their own work. And the demand curve only steepens from here: agents are starting to hold payment credentials and act unsupervised, as we covered in China's payment industry adopts a 'Know Your Agent' rulebook for AI that spends money. At that point, independent adversarial testing stops being optional.
Driverless middle-mile trucker Gatik raised a $200 million Series D led by Qatar Investment Authority and Koch Disruptive Technologies to push its fleet past 100 trucks by year-end. The company runs short regional routes from distribution centers to retail stores, and PepsiCo alone has put 41 autonomous trucks on Frito-Lay runs around Dallas, Phoenix and northwest Arkansas since signing a multiyear agreement in June. According to Gatik, it has completed 85,000 fully driverless orders with 99% arriving on time, contracted revenue tops $600 million across customers including Walmart, Kroger, Tyson Foods and Loblaw, and dynamic routing is stretching the early sub-10-mile fixed runs toward 400 miles. Total funding now sits around $500 million, with Cathie Wood's ARK among the returning backers.
The robotaxi race gets the headlines, but middle-mile retail freight is quietly becoming autonomy's most bankable niche: repeat routes, commercial customers who care about cost per pallet, and no nervous passengers to reassure. We tracked the heavy-truck end of this shift when Pony.ai's fourth-gen driverless heavy truck enters mass production — Gatik's numbers suggest the light-duty version of that thesis is already paying for itself.
What to watch: whether Alice's growth survives labs building red teams in-house — and whether Gatik's 100-truck milestone converts into new customer logos beyond its big-retail base.
If your job was to break an AI model before someone else did, would you rather work in-house or at a firm like Alice? Tell us in the comments.
