Securing AI: state of the market at the beginning of Q3 2023

Earlier in April, I published a v1 of the AI security ecosystem map focusing on companies whose mission is to secure AI and ML systems. 

Source: VentureBeat

A lot has changed since April. For once, we are hearing about more advancements in AI, and it feels like everyone is trying to build a company using LLMs. Moreover, we’re slowly starting to realize that while AI presents a great opportunity to transform the enterprise and greatly improve productivity, it also introduces a broad range of risks, most of which are not yet understood. 

In the context of security, most discussions about AI are still centered around questions “How can we leverage AI to defend our infrastructure?” and “What are the ways in which bad actors can achieve their goals by using AI and LLM?”. While that’s critical, we must not forget that someone must ensure that AI and ML workloads themselves are robust and secure. The number of players tackling these issues is starting to grow - it now includes the government, not-for-profits and community initiatives, and, naturally, passionate entrepreneurs. 

Given the speed of changes, it’s time for the updated market map of the AI/ML Security Landscape.

AI/ML Security Landscape at the beginning of Q3 2023

The below map focuses on AI/ML security solutions. It doesn’t include: 

• Compliance and governance solutions for AI and ML

• Tools that solve problems of privacy 

• Cybersecurity products that leverage LLMs to do what they are doing

The methodology for this chart is simple and meant to exclude bias. On the X axis is the amount of money raised, on they Y axis we plotted LinkedIn followers. While neither of these are true signs of business success, they have the advantage of being metrics that are correlated traction in the market and companies tend to focus on them as a natural part of running a startup.

It’s easy to observe that despite the market downturn, companies in AI and ML security have seen visible interest from investors. In the past quarter, 

• CalypsoAI raised a $23M Series A round led by Paladin Capital Group with participation from Lockheed Martin Ventures, Hakluyt Capital, and Expeditions Fund.

• KPMG announced the spin-out of Cranium, an AI security company developed within KPMG Studio, KPMG’s startup incubator,  and in collaboration with the firm's AI security experts.

Note that this market map is most definitely incomplete - many startups are operating in stealth, and more are emerging daily. Most of the companies on the chart (with notable exceptions of Cadea and Cranium) were started before 2023; it is anyone’s guess how many AI and ML security solutions are being incubated right now. 

The state of AI and LLM security will likely be drastically different just six months from now. Moreover, it is too early to speculate who has the potential to become the winner in this space. The best we can do is evaluate objective metrics visible externally, such as capital raised, number of employees, and social media followers, look at the track record of entrepreneurs building new companies, and the media coverage they get.