Join us in returning to NYC on June 5th to collaborate with executive leaders in exploring comprehensive methods for auditing AI models regarding bias, performance, and ethical compliance across diverse organizations. Find out how you can attend here.
AutoAlign, a leading provider of AI security solutions, announced on Tuesday that it has spun off from its parent company Armilla to focus on protecting and enhancing AI systems for enterprise clients. The company also unveiled its flagship product, “Sidecar,” a new approach to AI security that works alongside existing models to detect and mitigate potential issues such as data hallucinations, jailbreaking attempts, and biases.
The spin-off comes as Armilla shifts its focus to providing risk transfer solutions for AI systems to the insurance industry. “There’s actually an inherent conflict if we’re on the one hand, protecting [AI systems] and making them more robust, and then going and insuring them,” Dan Adamson, AutoAlign’s interim CEO and co-founder, said in an interview with VentureBeat. “Both parts of this are going gangbusters…but it’s really important that they’re done as separate entities.”
Addressing the growing need for AI security
As more enterprises look to adopt AI technologies, the need for robust security solutions has become increasingly apparent. “We’re seeing a lot of prototyping. But we’re not seeing a lot of these go into production because of these issues,” Adamson said. “It’s a huge problem…over three fourths of [CIOs and CTOs] have as their primary concern, the risks around [launching AI systems] as the number one reason why they’re not bringing them out into into production.”
AutoAlign’s Sidecar platform aims to address these concerns by providing a contextual, adaptive approach to AI security. “You can keep or preserve the power of these stronger models, while making them safe and effective,” Adamson told VentureBeat.
VB Event
The AI Impact Tour: The AI Audit
Join us as we return to NYC on June 5th to engage with top executive leaders, delving into strategies for auditing AI models to ensure fairness, optimal performance, and ethical compliance across diverse organizations. Secure your attendance for this exclusive invite-only event.
Request an invite
Operationalizing responsible AI with KPMG partnership
AutoAlign also announced a cooperation agreement with KPMG Canada to map the consulting giant’s trustworthy AI policy framework to actual alignment controls in AutoAlign’s system. “KPMG has been a leader for a while now around responsible AI, and they have a trustworthy AI framework that’s very much at the policy level,” Adamson said. “And so what we have done with this agreement is we’re able to now map those policies to actual alignment controls within Autoalign, so they can map each principle to these controls.”
This partnership aims to help KPMG Canada’s clients operationalize responsible AI practices and accelerate the deployment of AI systems to production. “For their clients, they can map these controls once, and then keep that as a Sidecar. And as their models evolve, or if they reuse or build on different use cases, they can keep that Sidecar that corresponds to those policies and preserve that,” Adamson said.
Spinout positions AutoAlign for focus
During our interview, Adamson also demonstrated how the Sidecar platform can nudge models to provide factual, referenced answers on sensitive topics rather than refusing to engage, a common limitation of other AI security approaches. “You can think of other use cases where we have technical documentation around sensitive subjects, like medical devices, where the models are like, I don’t feel comfortable answering about medical devices,” he explained.
This capability could prove crucial for enterprises looking to leverage AI in domains such as healthcare, finance, and politics, where accuracy and context are paramount. By enabling models to provide nuanced, contextual responses, AutoAlign’s Sidecar platform could help businesses navigate the complex ethical and security challenges associated with AI adoption.
Poised for impact in the enterprise AI security market
With the backing of Armilla and a partnership with KPMG already in place, AutoAlign is well-positioned to make a significant impact in the enterprise AI security market. The company’s unique approach, which focuses on enhancing rather than restricting AI models, could prove to be a key differentiator in a rapidly evolving landscape.
As the capabilities of foundation models continue to advance at an unprecedented pace, solutions like Sidecar will likely become increasingly essential for enterprises seeking to harness the power of AI while ensuring the responsible and safe deployment of these systems. AutoAlign’s launch marks an important milestone in the development of AI security technologies, paving the way for greater adoption and innovation across industries.