AI adoption is increasing faster than awareness of its security, according to predictions by Claire Lebarz, CTO at Malt.
“2026 will be the year of AI security,” begins Claire Lebarz, Chief Technology Officer at Malt. Malt is a platform for managing freelancers in Europe. The number of AI projects on Malt in 2024 has increased by 230 percent, demonstrating that organizations are interested in scaling AI.
According to her, the growth in AI adoption also increases the attack surface, including through prompt injections and less controlled actions by AI agents. At the same time, Lebarz expects that 2026 will mainly be characterized by experimentation: with safer agents and with technical teams that are reorganizing around new AI roles.
“Malt currently has 90,000 clients looking for independent talent, and 900,000 freelancers active on the platform,” she says. In Belgium, there are approximately 27,000 active freelancers.
Adoption overtakes security
“Over the past year, AI adoption has increased much faster than awareness of security,” Lebarz knows. According to her, this rapid scaling simultaneously increases the attack surface, from injecting prompts into AI browsers to uncontrolled actions by agents.
read also
Belgian companies embrace AI, but oversight is lacking
Although cybersecurity projects have increased by 35 percent according to Malt’s figures, insufficient investment is still being made. But according to her, the real challenge lies in the security considerations that must be integrated into the development of AI: “Governance, compliance and model behavior must be managed in the same way as the quality of code or data today,” says Lebarz.
Early stages
AI agents exist today in various forms. “The frameworks for AI agents are attracting a lot of attention, but their application within companies remains limited,” Lebarz notes. According to her, this is because companies must align the cloud, security and automation of workflows to realize AI agents.
The adoption of AI agents is still in its early stages: 2026 will be another year of experimentation and increasing adoption.
Claire Lebarz, Chief Technology Officer at Malt
“In practice, AI requires architectural maturity and at the moment we are still in a learning phase. Most use cases in companies remain based on RAG (Recovery Augmented Generation) and workflow automation rather than on autonomous systems,” she knows.
According to her, we are heading towards a year of experimenting and learning to make agents reliable, verifiable and safe before they will be effectively used in production.
Technical teams are reorganizing
Lebarz also notes a change in the way teams of engineers are structured. “Companies are looking for versatile profiles that can bridge the gap between the product, data and infrastructure, as well as experts in the field of security and AI,” she says.
We have seen a strong demand for new roles such as AI engineers, AI operations and AI project managers.
Claire Lebarz, Chief Technology Officer at Malt
Furthermore, Lebarz predicts that the old job description “backend only” will likely disappear quickly. She expects that the organization of teams will change, with fewer hierarchical levels and more autonomy and smaller full-stack teams, which include data skills from the start.
“We will very likely see more innovations in the way technical and product teams are organized,” she concludes.