Spain’s securities markets supervisor, the National Securities Market Commission (CNMV), will launch an ambitious technology implementation plan next year that includes the use of artificial intelligence tools in several areas, linked to the risk of market manipulation, false recommendations, financial scams and misleading advertising. These initiatives are part of a four-year digital transformation project launched by the new management of the Supervisory Authority, called Helix, and which will involve the investment of 24 million euros over this period.
“Artificial intelligence is already used for scams and favors or to facilitate this type of behavior,” says Asunción Gilsanz, director of the CNMV’s Technology and Digital Innovation department since May. “We don’t know if with Helix we will be able to redress the balance, but what is certain is that our technicians and experts will have advanced tools that will help them to better and better identify possible market abuse and fraud behaviors.”
The supervisor chaired by Carlos San Basilio predicts that the first applications of artificial intelligence, the first use cases, will be operational as early as 2026, according to a document to which this newspaper had access. Specifically, and in addition to the automation of internal processes, the CNMV hopes to use machine learning for the identification of market abuse, based on the information with which the organization already works. “With the application of machine learning, “Analyzing large amounts of data will be more effective and help identify patterns that identify possible fraud,” adds Gilsanz. Generative artificial intelligence will also allow, according to the CNMV, to extract data, verify and interpret official information sent by entities to the supervisor and keep track of historical databases, with the aim of detecting “misleading or incomplete or incorrect information”.
In addition to the usual market information, the supervisory authority will also use the new tools to monitor what happens on the Internet and social networks. It will then use artificial intelligence to scan the web and automatically identify financial scams, i.e. companies offering investment services without being authorized to do so. Fraudulent schemes that use technology and networks to reach more victims at very low cost. The CNMV will also use these mechanisms to track recommendations and possible relevant information that have not been disclosed publicly by listed companies. The plan involves using automated models to detect possible financial stress situations, but it won’t be the primary use, according to agency sources.
The initiative does not arise from nothing: already in the 2025 activity plan the CNMV had announced that it wanted to deepen the use of this technology both in internal processes and in new use cases. “Analytical artificial intelligence will be applied in supervision, optimizing times and expanding tools, in order to increase the ability to detect irregularities and risks,” he noted. This business plan already included an analysis of algorithms that use artificial intelligence and can pose a risk to the market.
On the employment front, in 2024 the CNMV launched the largest job offer in its history, with the call for 81 positions, with a significant weight of technological profiles. 54 people work in the technological department and another 14 technicians will become part of the next public tender for employment. “The plan is designed so that the entire workforce is part of Helix,” says Gilsanz. The project also involves a slight internal restructuring: the name of the Information Systems department is changed to Digital Technology and Innovation, a sub-directorate for digital transformation and an office to coordinate the different projects are created.
The use of AI is common in listed companies, market entities and even by malicious agents, and is receiving growing interest from the world of supervision, an environment in which this type of digital supervision tools are called suptech. “Tools driven by artificial intelligence allow us to process large amounts of complex data, from transaction logs to opinions on social networks, in order to detect anomalies and emerging risks. The use of large language models allows us to transform unstructured textual data into meaningful risk indicators,” explained the president of the Belgian supervisory authority at an Iosco (global association of securities supervisors) conference. The US SEC announced on August 1 this year (before its activities were reduced to a minimum due to the federal shutdown) the launch of a task force for the development of AI. This group will “equip SEC staff with AI-powered tools and systems to responsibly increase staff capacity, accelerate innovation, and improve efficiency and accuracy,” said Chairman Paul Atkins.
“New automated tools are increasingly used by market participants in areas such as fraud detection, regulatory reporting and risk management, while potential applications of new tools for regulators include increased surveillance capacity and improved data collection and management,” explains ESMA, supervisor of European markets.
