Alice Raises $140M to Secure AI Models

Alice, a company focused on securing AI models, has announced a significant $140 million funding round led by Apex Digital. This capital injection will fuel the company's research into ensuring that artificial intelligence operates as intended and remains secure against misuse. In an interview, Noam Schwartz, CEO and Co-Founder of Alice, detailed the company's origins and its relevance in the current AI landscape, especially following recent high-profile incidents involving advanced AI models.

The Genesis of Alice's Mission

Alice's journey began eight years ago, long before AI became a mainstream concern. The company initially focused on safeguarding social media platforms and major tech companies against a broad spectrum of online threats, including fraud, abuse, manipulation, and cyberattacks. "If you use the internet today, you probably were protected by something that we created," Schwartz stated, highlighting the company's extensive experience in identifying and mitigating harmful online activities. This foundational knowledge in combating online threats has been crucial in adapting to the evolving AI landscape. As AI capabilities expand, so do the potential avenues for malicious use, amplifying the harms that Alice has long worked to prevent.

The full discussion can be found on Bloomberg Technology's YouTube channel.

Addressing AI's Evolving Risks

The recent events, where two OpenAI frontier models reportedly escaped a sandbox environment and breached a third-party platform, have underscored the urgency of AI safety. Schwartz acknowledged that while this specific incident was attributed to human error rather than the models themselves "escaping," it highlights the increasing capabilities of AI and the potential for sophisticated harms. "Models are becoming more and more capable in coding, in cybersecurity, in other topics, and they would be even more capable in creating harms around different topics," he explained. Alice's core mission is to remain at the forefront of discovering potential risks, ensuring that its customers, which include major AI developers and Fortune 500 companies, can confidently advance AI without fear of unintended consequences.

Stress-Testing and Securing AI

Alice's approach involves rigorous stress-testing of AI models to identify vulnerabilities and potential failure points. The company works closely with leading AI developers, such as Anthropic, to "stress test models to figure out what could happen and break them and see that they're actually getting broken and provide the solutions for it." This proactive strategy aims to train models to be safer by identifying and rectifying potential issues before they can be exploited. Alice also works with enterprises that utilize AI products or open-weight models, ensuring they achieve a level of safety comparable to the industry's most advanced models. Schwartz emphasized the growing demand for AI democratization, stating, "Right now, there's democratization of harm as well when we want to make sure there's also democratization of protection and safety and security."

Strategic Funding for Future Growth

The $140 million raised from investors like Apex Digital, along with participation from SentinelOne and Samsung, will be instrumental in scaling Alice's research and development efforts. This funding will allow the company to continue its work in discovering potential AI risks, thereby enabling its clients to deploy AI technologies more safely and effectively. Alice's comprehensive approach, which includes pre-launch stress testing and ongoing monitoring, aims to provide a robust safety net for the rapidly expanding field of artificial intelligence.