The AI Industry’s Security Challenge: Is Collaboration Possible?

COMMENTARY: The security risks around AI aren’t just about new attack types like prompt injection or data poisoning – they’re about the fact that most teams are building in isolation, with no shared playbook. And that’s a problem. Right now, collaboration feels like a nice idea no one actually wants to act on. Too much competition, too little trust. Everyone’s racing to out-innovate the next guy, and sharing what works – especially when it comes to security – feels like giving away leverage.But if you’ve spent any time in this space, you know how fast the same mistakes keep resurfacing. Leaky APIs. Poor access controls. Misconfigured models. The AI world isn’t short on talent – it’s short on alignment. And while we probably won’t see industry-wide collaboration anytime soon, smaller, scoped efforts between trusted players are worth pushing for. MSSPs especially can’t afford to wait. We need to get smarter about where and how we share, and stop pretending we’ll all be fine figuring it out on our own.The rise of AI presents a new frontier for managed security service providers (MSSPs). It requires understanding a host of new attack vectors — adversarial attacks, model inversion attacks, data poisoning, prompt injections — and developing defenses that can withstand them.Artificial intelligence also presents a number of internal security vulnerabilities that must be kept in check, as the recent DeepSeek failure revealed. To keep AI systems secure throughout the entire development and deployment lifecycle, MSSPs must be able to address the unique vulnerabilities found in AI databases and models.One of the biggest challenges MSSPs must contend with in AI is a lack of standardized security practices and tools. The rapidly evolving landscape, limited expertise among security professionals, and challenges related to integration with other systems all contribute to the lack of standardization.However, the biggest hurdle to security standardization is AI developers’ reluctance to collaborate on developing effective solutions. Collaboration has the potential to streamline the development of new standards by bringing together industry leaders to share their experiences and foster the type of synergy that is often needed to develop novel solutions.Collaboration allows companies to contribute their strengths to improve the broader ecosystem. But inspiring it in AI means overcoming significant operational barriers.

Understanding the hurdles to AI security collaboration

Market competition is a perennial deterrent to collaboration. Companies seeking to gain competitive advantage closely guard their experience, insights, and innovations. Sharing them with competitors dilutes their value and diminishes the return on the investment companies initially put into acquiring them.Because AI is a fledgling and fast-growing industry, competition is even more intense. Companies that stay ahead of the curve attract the attention of consumers, investors, and talent. Consequently, AI companies can be reluctant to collaborate, even in areas as crucial as security, where shared progress would benefit the entire industry.Getting AI companies to collaborate on security issues can be a particularly tough sell, especially considering that data security is a core component of successful AI development. Companies that develop uniquely effective security tools and practices gain a valuable edge, but sharing their insights levels the playing field and erases that edge.Collaboration hinges on mutual trust – something still emerging in this young, fast-moving industry. Many AI companies are very new – for example, AI giant OpenAI is only a decade old – and haven’t had many opportunities to prove themselves trustworthy. Without a foundation of trust, collaborative efforts will struggle to reach a level at which valuable information is shared so that valuable solutions can be developed.

Taking steps to foster AI security collaboration

Companies that want to foster collaboration must start by building relationships. As those relationships are pursued, AI companies can identify other businesses open to sharing their expertise and experience to improve overall industry security. They can also begin to discern who can contribute something valuable and be trusted to engage with integrity.Additionally, as those relationships evolve, collaboration can become more formal by establishing a clear scope for the work that needs to be done. As needs and goals are more clearly defined, companies may find that others are more willing to enter the discussion. Without clarity, those invited to collaborate won’t know what they are risking or how their contributions could ultimately be used.Although collaboration can streamline the development of the security standards needed to guide MSSPs in the rapidly evolving AI space, several key challenges must be overcome before they can be leveraged optimally. By building stronger relationships and adding clarity and structure to the collaboration process, AI companies can encourage their colleagues to join them in seeking out much-needed security solutions.